Hoffman Amplifiers Tube Amplifier Forum
Amp Stuff => Tube Amp Building - Tweaks - Repairs => Topic started by: wittyjeff on May 15, 2021, 08:39:14 pm
-
I have attached a schematic that lists my voltage measurements (in orange) on a new build that is like a 50W Plexi power amp with a "Ultra High Gain" preamp from Merlin's Designing Preamps book on page 260. I've listed my questions and comments on the attached schematic png file. Any helpful insights would be appreciated. Thanks in advance.
-
Hey mate, All your voltages look OK for that circuit.
As for running DC heaters...
Put the 6.3v through a voltage doubler into a 12v regulator and wire your three preamp valves to take 12v (pins4,5)
-
-36V bias seems absurdly high for EL84 at 350V. Did you check the datasheet?
-
Well done in making an ambitious build a reality!
Point #7 is the only query that seems particularly odd, as there should be 11VDC across the 4k7. Does the presence control work, are you sure that node isn’t getting shorted to 0V common somewhere?
As general observations, is the VVR to the output valves maxxed out, ie you’re never intending to go above 350V?
And the VVR isn’t shown as being applied to the bias feed, surely that’s not going to work out well when the HT feed gets lowered?
Output valve cathode current ie mV measurements are really needed to judge whether the apparent anode and screen grid current draws seem normal. eg 1mA idle screen grid current seems lower than normal, but need the anode or cathode current to put it in context.
-
@ PRR
On the schematic I read EL34
Franco
-
Some more thoughts:
The voltage to the PI seems extremely high. I would drop voltages a bit before the PI.
Why increase the PI plate resistor values? It should mess with the bias of the whole PI. In addition, the difference between the PI plate resistors is now a lot bigger than with the standard 82k/100k, which could increse imbalance. Is that intended?
-
Good point, why increase R7 (to 30k over the typical 10k) rather than R85.
To mitigate for excessive bias shift (leading to blocking distortion) when the power amp gets overdriven, it’s best not to have a lot more max potential peak voltage swing available at the preceding stage’s output than the magnitude of the output valve’s bias voltage.
Hence perhaps Marshall’s choice of 20k HT dropper to the LTP node, for many EL34 models, following the 8k2 dropper of the KT66 JTM45 models.
-
Hey mate, All your voltages look OK for that circuit.
As for running DC heaters...
Put the 6.3v through a voltage doubler into a 12v regulator and wire your three preamp valves to take 12v (pins4,5)
TIMBO, Thanks! great idea on a more efficient DC heater supply. I just grabbed an off the shelf DC supply board that seemed like it really smoothed out the DC well (http://www.pmillett.com/DC_filament_supply.htm). I think I'll go the 12V route on the next build for sure.
-
PDF64: As always. Thanks for your help and input. My builds are way more ambitious than my abilities... but that's how I learn (mostly through embarrassing myself and making mistakes).
Point #7 is the only query that seems particularly odd, as there should be 11VDC across the 4k7. Does the presence control work, are you sure that node isn’t getting shorted to 0V common somewhere?
It is odd indeed. Based on your confirmation, I think we definitely have a problem here. I'll hunt for the unintentional grounding. I'm not far enough along (wanted to get the voltages correct first) to see if the presence control works, although I did plug in a guitar to see if I got sound. Something is jacked-up since when either the gain or volume is turned up past 20%, it makes a weird noise. I'll eventually track that down, but I'll get the voltages set up correctly first and figure out where that 11VDC went.
As general observations, is the VVR to the output valves maxxed out, ie you’re never intending to go above 350V?
And the VVR isn’t shown as being applied to the bias feed, surely that’s not going to work out well when the HT feed gets lowered?
My purpose for installing a VVR on this build is to facilitate the option to install different Power Tubes. I'm not using the VVR as a type of Master Volume. The Voltage Adjustment Knob is on the back and whenever a tube set is changed, the bias will need to be adjusted manually as normal. I figured that way one can dial in exactly what bias is needed for the tone sought and tube set.
Right now the knob is at about 40% but the max voltage is around 475VDC. I'm hoping to be able to run some KT88s in here (barely) as well as try some 6V6s. I've increased the range of the bias supply and installed a 10-Turn Bias Voltage Adjustment pot. All of which are accessed from the exterior (including voltage test points).
Output valve cathode current ie mV measurements are really needed to judge whether the apparent anode and screen grid current draws seem normal. eg 1mA idle screen grid current seems lower than normal, but need the anode or cathode current to put it in context.
Got it. From this and the other comments I'll reply to next, it's apparent that I forgot/don't understand the relationship of current and voltage drop. I think I'm getting it a little more clearly now. I'll check these out once I make the other changes I need.
-
Some more thoughts:
The voltage to the PI seems extremely high. I would drop voltages a bit before the PI.
Why increase the PI plate resistor values? It should mess with the bias of the whole PI. In addition, the difference between the PI plate resistors is now a lot bigger than with the standard 82k/100k, which could increse imbalance. Is that intended?
This is just a blatant newbie lack of understanding on my part. Embarrassing as it is now that you guys have enlightened me and now I'm recalling stuff I've read regarding this... I was just thinking, hmmm... I need lower voltages on these plates, it's too much work to tear apart some stuff to change that R85 resistor, so I'll just slap some larger resistors on those plate supplies. Stupid, ...now I know.
-
Good point, why increase R7 (to 30k over the typical 10k) rather than R85.
To mitigate for excessive bias shift (leading to blocking distortion) when the power amp gets overdriven, it’s best not to have a lot more max potential peak voltage swing available at the preceding stage’s output than the magnitude of the output valve’s bias voltage.
Hence perhaps Marshall’s choice of 20k HT dropper to the LTP node, for many EL34 models, following the 8k2 dropper of the KT66 JTM45 models.
As I mentioned in my previous comment, the reason I put the 30K where I did was just a lack of understanding and laziness on my part. I'll have to change the 10K R85 to a 20K and change the R7 back to a 10K and see where I'm at (also put back the 82K and 100K plate resistors).
However... if you'll note that I also have a R87 at 10K between the last stage of the pre-amp and the phase invertor (a total of 40K!)... that's not good (now that you have enlightened/reminded me). I'm thinking that I should eliminate the R7 entirely so that there's only a 10K between the stages and then put a heftier resistance (perhaps 30K?) on R85... am I thinking more correctly now?
-
Some more thoughts:
The voltage to the PI seems extremely high. I would drop voltages a bit before the PI.
Why increase the PI plate resistor values? It should mess with the bias of the whole PI. In addition, the difference between the PI plate resistors is now a lot bigger than with the standard 82k/100k, which could increse imbalance. Is that intended?
This is just a blatant newbie lack of understanding on my part. Embarrassing as it is now that you guys have enlightened me and now I'm recalling stuff I've read regarding this... I was just thinking, hmmm... I need lower voltages on these plates, it's too much work to tear apart some stuff to change that R85 resistor, so I'll just slap some larger resistors on those plate supplies. Stupid, ...now I know.
Nah, the only stupid questions are the ones not asked. You came here for advice, and I can see that you're learning and making the best of it. That's what this type of forum is for.
Good luck!
-
Ok, I put back the correct resistors to the plates of the PI (the 82K and 100K) and I moved the voltage dropping resistor R7 to be BEFORE the whole preamp. The reason I didn't just change R85 is that it's kinda buried. I'm assuming that now R7 should allow the preamp to function correctly. I ended up needing to use 24K in order to get the voltages low enough (they are low enough, right?) I've attached a revised schematic with the updated voltage readings that I took for this part of the process. Let me know if the voltages are now ok or if more fine-tuning is needed.
I turned the bias knob so that the bias voltage is now -30V, I guess the -36V was making some nervous. I have not even checked the bias status yet, so I'm not sure where it will end up for this EL34 tube set. Again, remember that the HT supply to the power section is adjustable so it can be set wherever. Previously I did mis-state that the Voltage Adjustment Knob was turned at 40%. Well, apparently I installed it backward, so it's really at 60%. Either way, there's some upward ability (to around 475V) yet and of course there's plenty of downward range. Again, this is just to accommodate different power tube sets (with a corresponding manual bias adj when changing tube sets).
-
I figured out why there was no voltage on the PI tail (or whatever it's called). Instead of a 4K7 resistor I installed a 47R resistor for some reason. Now there's 7.8VDC just prior to that 4K7 resistor (R51). *see attached
I know household voltage goes up and down a bit so each time I take measurements, even when nothing is changed, there are small differences (I'm assuming since the AC unit is running and so is the dryer, dishwasher, etc). Anyway, with the supply to the PI up 3 volts (from 299 to 302), the plate voltage when from 239V to 246V on one side and from 194 to 205V on the other. That's quite a jump. Apparently fixing that resistor (R51) affected the plate voltages too... I guess it makes sense with more resistance from cathode to ground.
Is the 246V and 205V too high?
-
On the schematic I read EL34
My little dog is now half-blind and it seems to have rubbed off on me. :embarrassed:
-
Just checking in to see if the voltages in the PI area look ok. (see above)
Also, I measured the current from the OT center tap to the Power tube plates and it's very low. For example: Plate Voltage (pin 8) 302.7VDC, OT Center Tap Voltage: 303.2VDC and the current between them is only 0.06mA. Ideas of what's going on or where to start looking?
-
Maybe the output valves are in cutoff?
Not got time to plot it out from the anode’s triode characteristics at the moment.
Try increasing the magnitudes of the screen grid voltage / decreasing the magnitude of the control grid voltage.
-
Maybe the output valves are in cutoff?
Not got time to plot it out from the anode’s triode characteristics at the moment.
Try increasing the magnitudes of the screen grid voltage / decreasing the magnitude of the control grid voltage.
You may be right. From the Tube data sheet, I see that at 300V Plate voltage and a control grid voltage of -30V, the current would be very low. I tried to lower the control grid voltage but could only get it down to -24V which did bring the current up from 0.06mA to 0.3mA ...so there's that.
On a typical Marshall type Power Amp running two EL34s, I thought I saw a typical control grid voltage of around -30V but I could be wrong. Isn't a Plate Voltage of around 300V typical as well? I can dial the plate (the Power Amp B+) to whatever and I can set the bias to whatever (although it's not going below -24V but I can change that since it appears to be able to dial up to more than -90V, so I have plenty of range). ...but what should the "whatevers" really be? Since there's barely any current flowing, the Screens basically are reading at nearly the same as the plate voltages, not sure how/if I can change the screen voltages.
Thoughts?
-
I think typical Marshall 50W HT ranges from 400 to 470V.
See https://el34world.com/charts/Schematics/files/Marshall/Marshall_ampchart.gif
But bear in mind that the analogue meter used (eg 20kohms / volt) may have loaded the control grid voltages down a bit.
The screen grid voltage can be raised (thereby bringing the anode current up out of cutoff) by turning up the VVR.
From the Tube data sheet, I see that at 300V Plate voltage and a control grid voltage of -30V, the current would be very low. I tried to lower the control grid voltage but could only get it down to -24V which did bring the current up from 0.06mA to 0.3mA
Ok, steady on, the triode curves on p9 of https://tubedata.altanatubes.com.br/sheets/129/e/EL34.pdf
indicate anode current should be well above that.
I suggest to remove the output valves and check the pins 1,3,4,5,8VDC.
-
I think typical Marshall 50W HT ranges from 400 to 470V.
See https://el34world.com/charts/Schematics/files/Marshall/Marshall_ampchart.gif
But bear in mind that the analogue meter used (eg 20kohms / volt) may have loaded the control grid voltages down a bit.
The screen grid voltage can be raised (thereby bringing the anode current up out of cutoff) by turning up the VVR.
Thanks for the voltage charts. Good info. So according to the chart a 50W Marshall in general should have a pin 3 (plate) voltage of 395V and a pin 4 (screen) voltage of 390V and a pin 5 (control grid) voltage of -44V. So... I adjusted things to get 395.5V on pin 3 which resulted in a pin 4 (screen) voltage of 393V and I set it to a pin 5 (control grid) voltage of -43.8V. After doing that, the current from plate (pin 3) to the CT of the OT was 0.1mA on one tube and 0.04mA on the other tube. What other readings should I take? ...a plate to cathode current?
I don't think it should be that low, right? Also, why those tubes would be different from each other is another new mystery, they are new tubes, matched, and the bias (all inputs for that matter) is exactly the same (well within a decimal or two, I have a bias balance pot).
Thoughts?
-
Ok, steady on, the triode curves on p9 of https://tubedata.altanatubes.com.br/sheets/129/e/EL34.pdf
indicate anode current should be well above that.
I suggest to remove the output valves and check the pins 1,3,4,5,8VDC.
Ok. Reading this now...
-
Just a quick strange note... first, I noticed that in my reply to pdf64 above, I used the voltages for the 6550 and not the EL34 from the 50W Marshall chart, so I proceeded to repeat the measurements but setting things to the correct voltages for the EL34 set-up, (i.e. 380V plate voltage and -30V control grid voltage)... I noticed that after adjusting the bias current so that the DC on the control grids was -30V, the plate voltage went down considerably. So... no big deal I adjusted the VVR to maximum. Even after "settling" for awhile the maximum supply from the VVR to the Power Section was only 342V. Which is weird because this thing ought to be able to put out well over 400V even under load. For some reason I messed with the Solid State/Tube rectification switch (which I've done before, they both seem to perform as expected). It seems like the process of switching did something to the VVR to kinda "reset" it since then it started to allow the voltage to increase up past 400+V again (I had the knob set at max trying to get it over the 342V mark). I thought, maybe the Solid State rectifier is whacked, but after switching the rectification back and forth, the VVR/Voltage level still stayed up there regardless of whether I had it set to Solid State or tube (I have a GZ34 in there now).
It gets stranger. While in this "new condition" (i.e. VVR seeming to now be allowing the expected higher voltage) I then set the voltage to 380V for the plate voltage (still have the control grid at -30V). When I took a Screen voltage reading, the voltage was much lower than the plate voltage. This was very exiting since this is what I would expect and what was mystifying me as to why the screen was always showing almost the exact same voltage at the plate (which I think indicates a low current flow if I'm understanding things). My brain kinda kicked in when I realized that, with a large difference in voltage (I don't remember exactly how much, since I got kinda excited)... there must be alotof current flowing through those tubes! I quickly turned the lights down to look and sure enough... there was some red-plating already going on. I shut things down and went to type this.
So, before I fire things up again... any thoughts on what's going on and what the next step might be?
-
A control grid voltage of -30VDC measured with an old analogue 20k/V meter would probably measure somewhat different with a 10M input resistance DMM, eg -35V (just a guess, I’ve not calculated that).
Point being that by using those old voltage charts without correcting for that, you might be biasing it too hot.
Sorry, by not explaining that I kinda set you up for it :huh:
A key point is that when twiddling with the bias or HT voltages, always monitor the cathode current whilst doing so. The exact control grid voltage is secondary, what matters is how it affects the particular valves fitted.
-
A control grid voltage of -30VDC measured with an old analogue 20k/V meter would probably measure somewhat different with a 10M input resistance DMM, eg -35V (just a guess, I’ve not calculated that).
Point being that by using those old voltage charts without correcting for that, you might be biasing it too hot.
Sorry, by not explaining that I kinda set you up for it :huh:
A key point is that when twiddling with the bias or HT voltages, always monitor the cathode current whilst doing so. The exact control grid voltage is secondary, what matters is how it affects the particular valves fitted.
Got it. However, I've run the control grid voltage as low as -24 (so lower than what it was set at during the episode described above) and it still had little to no current flow. I'm saying that something has suddenly changed, probably something to do with the VVR circuit, whereby suddenly I'm getting a lot of current (albeit probably a correct amount for that particular bias setting at that moment).
So... it may be that things are "working" now as far as current flow through the power tubes. I'd like to see if I can duplicate the original problem exactly and find out the cause. I'll admit that I don't really understand the finer details of how the VVR circuit functions, but it looks simple enough.
So... here's a really stupid question, how best to measure cathode current? do I need to disconnect the cathode to ground and run it through the meter (there's a 1 Ohm resistor to ground on my schematic, FYI)? ...or can I just connect one lead on the cathode and one to ground. ...or are we talking plate to cathode current? Up to now I've been just using the directly measured plate to the CT of the OT current measurement (easy since I have those test points on the back and don't need to be like I'm playing that old "Operation" game if you know what I mean). ;)
-
By ‘monitor the cathode current’, I was thinking of clipping a meter (set to mVDC) across one of the 1ohm resistors.
-
The strangeness continues...
The not-so-strange part: So I turned the thing on again and set the bias voltage higher to -39.6 (at the control grid). With that bias setting and the plate voltage set to 385.9, the voltage at the CT of the OT was 387V. Using the voltage drop (of 1.1V) and the OT resistance (measured when warm but off through the plate to the CT of the OT) at 38.4 Ohms, I get around 28.65mA. If that's the case, then when you multiply that current by the plate voltage you get a plate dissipation of: 11.054. Divide that by 22W and you get around a 50% bias percentage. Did I do that right? So, in this scenario (at these settings at that particular moment), the power tubes seemed to be functioning within reason.
The strange part: After taking the measurements above and shutting things down for awhile, I turned the amp back on again. After settling in, the voltage measurements still were not as I left them. So that sucks. Again, I think the VVR isn't functioning quite as it should. Anyway, I re-adjusted the voltages to try to duplicate the above scenario. This time I couldn't get a reading with the same voltage drop.
For some reason I decided to plug in a speaker instead of a dummy load. With no input and with volumes all the way down, I noticed that as I turned the VVR, there was a spot where the voltage jumped up by 20Vs and at that spot and within roughly 10V of that spot either way, the speaker produced a buzzy (60hz-ish) sound. Not very loud, but from virtually NO noise to this buzz, but only when turning the VVR through this one limited range. I also did connect a guitar and it does send signal all the way through to speakers, but it is very mushed, not loud and as either volume is turned up, it adds this buzzy sound... so I'm guessing there's more than one thing jack-up here.
My thought as to a next move is, eliminate/bypass the VVR altogether. Of course I will need to replace the 10W 10 Ohm dropping resistor (R02) with something much higher to get the appropriate B+ voltage to the power section, but since that will also drop the pre-amp voltage I guess I'll just add a temporary resistor in somewhere in place of the VVR. Anyhow, with the VVR out of the circuit, I can simplify this thing and narrow down what's going on here.
Does that approach make sense? Any additional thoughts?
-
That seems a good plan.
I can’t see that R02 is dropping much voltage, but whatever, I’d just link across the VVR and send the 470VDC to the output valves.
Obviously need to turn the bias to its most negative value before powering up.
And for the initial power up, use an LBL just in case.
when you multiply that current by the plate voltage you get a plate dissipation of: 11.054. Divide that by 22W and you get around a 50% bias percentage
I don’t follow that :dontknow:
I suggest just to set the bias to get about 30mV across the 1ohm cathode current sensing resistors.
-
Got it. Will do and report back.
Thanks!
-
Ok. I bypassed the VVR. I'm assuming you meant monitor the voltage across one of the 1 ohm current sensing resistors in order to "monitor the cathode current." ...correct? Its a little confusing since, if I understand correctly, I believe we are actually looking at voltage in order to "monitor the current." Anyway, I first tested the start-up using my trusty lightbulb current limiter (that my wife always makes fun of) and things looked fine. After shutting it down and letting it cool off for awhile, I then clipped the Vmeter across that 1 ohm resistor and fired it up using normal household current (no LBL). I adjusted the bias knob until the Vmeter read 30mV. The other readings were: Plate voltage of 465.4V and a OT CT voltage of 466.3V and the bias was at -48V (which is where it ended up at in order to get the 30mV mentioned above).
-
Ok, all that seems good, the VVR can be addressed later.
Regarding the current - voltage thing, don’t get too fixated, we’re just using Ohm’s law and a sensitive volt meter to sense current.
Actually, it’s just what your meter is doing when it’s switched to measure current, ie there will a small current sensing resistor between the probe sockets, and it’s the voltage across that which is being measured.
The benefit of a 1ohm cathode resistor, rather than your meter in current mode, is that we don’t have to break the circuit, because the sensing resistor is already there.
Apologies if that has got you even more confused :icon_biggrin:
-
Thanks again for your help guiding this process. So where we're at now is... the VVR was acting squirly and likely causing a low current flow situation, at least at some points or intermittently. It also didn't seem to be stable since I would get it set and stabilized resulting in certain voltage readings... then when the amp was turned on again later, without touching any of the knobs/settings... the voltages would be quite different (even after waiting quite awhile for it to "settle"). So... goodness knows what is/was going on with that. I can try to deal with that later in a separate thread. With that out of the way, we can eliminate some of those shenanigans.
So now I've got a nice stable HT/B+ supply and the power tubes have the proper current flow (as indicated by the 30mV across the current sensing resistor) and the resulting bias is at 49% (plate dissipation of around 10-11W). So... I figure this thing should produce some sound now, right?
Well, it does produce sound but even with the gain and Master Volume only at 25% it is so distorted it's just mush and isn't able to get much volume. So... my uber gain is like way too much, or something is off.
I've rechecked the schematic compared to the layout and it seems like it's laid out correctly (it's not a very complicated circuit). ...it's possible some values are off such that the gain is so high in the early stages that it's all mush by the time it gets to the power section. Not sure where to start digging. I'm starting to think I should have used a tried and true original Marshall circuit since not many have built this Merlin circuit, but Merlin knows what he's doing so I figure I am safe to assume that instead of a design issue with the circuit... it's likely that I put in a wrong value component or mis-wired something. Any thoughts of how to proceed next?
-
A master vol amp is just a non master vol amp with another vol control that’s late in the signal chain, ie the master vol. If that master vol has an audio taper and is set to 50% rotation, its electrical output will typically be 1/10 of the input voltage, ie 20dB attenuation. Halve that again and turn it down to 25% rotation, and the electrical output might be 1/10 of a 1/10, ie 1/100, 40dB attenuation.
The point being, with the master vol turned down so much, the preamp signal has no chance of getting a decent signal to the power amp.
Hence I suggest that to check whether the amp can kick out a decent cleanish signal, turn the master vol up and the gain down.
-
Figure out how to get v1b's output to either the EQ section input or the PI's input, bypassing v2, and see how that sounds. Then do the same but from v2a. As pdf64 suggested, dime the MV so it's not a factor.
My experience with the Merlin medium gain pre was that it had way more distortion than I expected from his description. Impossible to get a good clean signal and even difficult to get a say late 60's/early 70's rock plexi sort of tone. And that was built on his modules. I would expect his "higher gain" designs to be even more so.
-
My experience with the Merlin medium gain pre was that it had way more distortion than I expected from his description.
Boy oh boy do I agree with that. Great writer otherwise, though.
-
Guys, as always, thanks for the great ideas and guidance. My day job (still working from home) ended up taking up my whole day so I wasn't able to implement these procedures just yet, but I'll try to get at it tonight or early tomorrow.
Hopefully we can track down which gain stage(s) are overdoing things. Otherwise, it wouldn't be all that terrible (just time-consuming and irritating to need to do it) to just solder up a different eyelet board based on a different circuit and slap it in there. My goal still is to create an amp that is able to dial in some heavy distortion (without a pedal). Again, hopefully we will be able to tame this Merlin beast into something useable. So... let's see if I can bypass V1 and V2 and see where the problem lies and go from there (great suggestion).
-
....If that master vol has an audio taper and is set to 50% rotation, its electrical output will typically be 1/10 of the input voltage, ie 20dB attenuation. Halve that again and turn it down to 25% rotation, and the electrical output might be 1/10 of a 1/10, ie 1/100, 40dB attenuation.....
Yes, 50% rotation on "Audio taper" is usually 20% to 10% electric output.
However 50% of 50% is not much less. It would be if we still had good "LOG" pots, but they vanished with analog computers and analog artillery aimers. The typical "Audio" today is two linear sections, joined around half-turn. If it really is 10% at mid-turn, then quarter-turn is likely to be 5% or 26dB.
Your point is still perfectly valid. "...with the master vol turned down so much, the preamp signal has no chance..." etc.
Although conversely, a high-level pre stage with a sensitive power stage, you may not be able to turn-down enough to strain the pre stage.
-
Just a quick side note that I should have stated earlier. I did turn both the gain and Master Vol to all possible combinations. The only sound this thing spits out is mush. There's no clean anywhere and the mush seems to be consistent throughout any possible gain or MVol pot settings. There may be multiple things going on since there's some 60 or 120hz buzz/hum in there too.
-
Ok, so I did a first step and bypassed V2 completely. So with only V1 feeding signal into the EQ section, (and with the Master Volume turned all the way up this time), it is VERY gainy. With the gain pot turned to maybe a 2, there is still some crunch on the notes if you attack the strings hard. Anything over 3 or 4 is just heavily distorted... like maybe what I would have settled for with the gain turned to like 9.
I did note that the 60/120Hz hum/buzz has disappeared in this configuration. Also, it's not very loud either no matter the settings.
Shouldn't the signal out of this V1 be mostly clean and just barely distorting only after the gain is turned past, say 8? Seems like the problem (or one of them anyway) is right here in V1. What's your best suggestion for calming these two stages down?
-
Have you got a scope, to check for oscillation?
-
Yes, I do. I just got one but haven't really used it yet. I'll try to read up and youtube how to check for oscillation etc and report back my findings. Good idea and makes sense for sure. (I get to do something useful with my new scope, yay!)
-
Ok, so here's my first oscilloscope measurement...
I attached probe 1 to the signal output of V1 but BEFORE the coupling capacitor prior to the eq. the 2nd probe was attached post-eq prior to the coupling capacitor into the PI. These two pictures (attached) are with both volumes at 100% and a guitar plugged in but not touched.
The first photo is with the voltage/div setting at 1 so each square = 1V. The second photo is with the voltage/div setting at .2 to amplify what's going on.
Basically, in this scenario, all I see is just noise from the guitar mostly. Is that what you're seeing?
-
My Second Oscilloscope reading: the same settings as above, but with probe 1 connected AFTER the coupling capacitor instead of before. The first two photos are the results with the guitar still plugged in but not touched. The second two photos are with the guitar unplugged. Without the guitar plugged in, most of the noise disappeared, which is to be expected. Regardless, I don't see anything that appears to be high frequency oscillations, do you?
I know that ideally I'd have a signal generator and run a sine wave through here, but I don't have one. I tried using the phone app signal generator, but that had plenty of it's own noise and distortion so I don't think it would lend much to the discussion, although I didn't try running it through the amp.
Thoughts?
-
The first photos below are with Probe 1 still in the same position (after the coupling cap to the eq, so showing only the signal amplification after V1), but this time with Probe 2 connected to the grid of one of the EL34s. No signal and with both gain and Master Volume at Maximum. The first photo shows the results but with the voltage/div at .2 (in order to amplify/exaggerate the noisy signal) and the second photo is with the voltage/div at 1. It looks pretty noisy, but I've actually had the speaker connected all of these times instead of the dummy load... there's almost no noise coming out of the speaker (just a little normal noise/buzz).
Thoughts?
-
None of that is very meaningful. You need to use a sig gen to inject a signal into the amp. And you need to connect your scope probe ground clip to the chassis.
-
I guess I'll need to buy a signal generator then. I thought I saw a video where the oscillation was visible even with no signal input, but I definitely see the need for a signal generator for this and several other uses... so I'll get one.
The ground clips were connected to the chassis/ground but I failed to mention that.
I guess let's see what a sine wave looks like once I get a sig gen.
-
I see now where the ground clip to the chassis comment came from. I can see in the photo there's a red alligator clip connected to the ground lug on the oscilloscope. That's just a spare alligator clip/wire that I was using for some other stuff and it's not connected to anything now. My probe ground clips are black. Good catch though... and totally something I might do.
-
Plug your guitar in and play a note. See what the scope shows then and let us see. Not as scientific as a sig gen but it will show you useful stuff. I have a sig gen app on my i-phone that I use, and also have ditto looper pedal I sometimes play a line into and then just let that repeat as needed so I'm hands free to fiddle with trimmers/clips/etc.
-
My ground clip comment was based solely on the scope display in your pics. You really don't need a sig gen to view the squeal. But you must set the amp such that you hear the squeal. Then connect a scope probe to the speaker jack so you can see the squeal. Set your vertical gain, timebase, and trigger to be able to see a stable oscillation (squeal) on the display. Now use your other scope probe to trace backward through the amp, looking for the point where the oscillation stops. You'll likely need to adjust the vertical gain of the second channel as you move from point to point in the amp, but don't touch the timebase or triggering! And use a speaker dummyload if you don't want to hear the squeal.
-
Plug your guitar in and play a note. See what the scope shows then and let us see. Not as scientific as a sig gen but it will show you useful stuff. I have a sig gen app on my i-phone that I use, and also have ditto looper pedal I sometimes play a line into and then just let that repeat as needed so I'm hands free to fiddle with trimmers/clips/etc.
While I'm waiting for my signal generator to arrive, I'll definitely try these and post the results. Great suggestions, thanks!
-
My ground clip comment was based solely on the scope display in your pics. You really don't need a sig gen to view the squeal. But you must set the amp such that you hear the squeal. Then connect a scope probe to the speaker jack so you can see the squeal. Set your vertical gain, timebase, and trigger to be able to see a stable oscillation (squeal) on the display. Now use your other scope probe to trace backward through the amp, looking for the point where the oscillation stops. You'll likely need to adjust the vertical gain of the second channel as you move from point to point in the amp, but don't touch the timebase or triggering! And use a speaker dummyload if you don't want to hear the squeal.
Ah, ok, good points. In this case there is no audible squeal at all. It's just that the signal is just heavily distorted mush and some suggested that there might be some parasitic oscillation (inaudible) that might be causing it. With no signal input, and gain and volume cranked, the amp is fairly quiet other than maybe a quieter-than-usual typical background hum/buzz. Even with V2 bypassed altogether, a guitar signal distorts when the gain is at 2 and is just mush at anything much beyond that. My thoughts are that V1 is just set up such that it is prematurely distorting. Still will need a good signal generator to see that more clearly on the oscilloscope and trace it back from the output, compare stages, etc. as you suggested.
-
When you connect a sig gen to the input set the level for 100mVPPand freq to about 500Hz as seen at the input jack with your scope. Then signal trace forward through the amp watching for the sine wave to start squaring off.
-
When you connect a sig gen to the input set the level for 100mVPPand freq to about 500Hz as seen at the input jack with your scope. Then signal trace forward through the amp watching for the sine wave to start squaring off.
Got it, will do. I did the stuff noted below before I saw this just now, I'll do this above suggestion next.
-
I checked out the signal generator app on my phone and through the amp it produces a reasonable sine wave. So, the first photo below shows what the signal looks like at the grid of an EL35 when both gain and volume are maxed. It has a nice sine wave but starts to distort about about 4.5. The Voltage/div was set on 5 so there's about a 5V swing on this signal here... but I didn't measure it at the actual input, so I suppose that info is useless. I'll be redoing all of this below according to Steve's suggestions, but thought I'd throw this out since I did it before reading his post.
The second photo is of the output as measured at the 4 Ohm speaker output tap, with all volumes at max. Note that the voltage/div is set to 5mV... so there's only about a 15-20mV swing on this output signal. Again, probably useless since I didn't measure the original input signal... but still, seems kinda low.
The third photo is of the same output signal, but with the gain turned down to about 5.
Anyway, for what it's worth, there's that. I'll try to get some readings that are a bit more useful next...
-
YOU DON'T HAVE TO RUN ALL POTS AT MAX.
-
Ok, here's another run at it...
Pict 1 is just of the input signal. Turns out I can only get a 50mV signal from my phone (it swings 100mV + to -, but I figured you wanted a 100mV amplitude, right?). Anyway, with that in mind...
Pict 2 is the signal after stage 1 (1st half of V1) beginning to distort with gain at less than 3, around 15V.
Pict 3 is the signal after stage 2 beginning to distort gain just less than 3, around 30V (probe set to x10)
Pict 4 is signal after stage 2 with gain cranked 100%, around 40V
Not sure if these are useful, but let me know what readings to take next. Interesting note, this time I was monitoring the voltage across the current sensing resistor from the power tube cathode and it got up to 50mV with things cranked.
-
Better looking pics.
-
I wonder if this isn't part of the problem...
It seems that the output of stage 1 (1st half of V1 as measured from the grid of stage 2) is clipping no matter what the gain knob is set to. Pict 1 attached is with the gain turned up a tiny bit (to around 1 ~5V), definitely clips here. The point is that this first stage seems to be clipping even at rather low gain settings. Anything over 3V is clipped. Note that this stage is only producing a max Voltage of under 20V with gain at max, this is with a 50mV input. How do we get it to stop clipping. In comparing this "Ultra High Gain" circuit with Merlin's "High Gain circuit, I see that the High Gain circuit has a higher plate voltage (a 100K plate resistor instead of a 220K) and a slightly lower value cathode resistor (which increases the cathode voltage from 1V to 1.4V). Do you think slapping in those changes would get rid of this first stage clipping?
-
The gain pot does not control the first stage.
You can change that 220K plate resistor to 100K to reduce clipping or you can reduce the signal level you are putting into the amp.
-
Are you sure it isn’t the 2nd stage grid clipping the signal, due to the positive peak exceeding the bias voltage?
eg if the signal level there is reduced a bit, does that reduce / eliminate the clipping?
-
I've connected the probe at the grid just prior to stage 2 as indicated by the red X on the attached schematic. Because I only have a phone app to produce the signal, it's only 50mV... so not equivalent to a 100mV guitar pickup signal. Regardless, at that signal input with the probe connected at that location, when the gain pot is turned to around 2, such that the signal at that location is around 3V, it's clipped (it's unclipped when the gain pot is below that)... I'm guessing being clipped like this is bad... so I need to fix it so such that it's not clipping at this early point. Am I thinking correctly on that?
What's causing this clipping and how to correct it is the part I'm not sure about.
My guess is that reducing that stage 1 plate resistor might help, but that's simply because I noticed that Merlin's similar but lesser gain pre-amp circuit uses a 100K resistor on the 1st stage plate (and because Steve mentioned it as well). I can't really reduce the input signal any more, it's already at half of what a guitar pickup would be sending (well, I guess I could, but I don't know how that would result in beneficial information). Hopefully I'll get the signal generator from Amazon tomorrow and presumably it will be adjustable to whatever signal strength I set it at, so we can see what a 100mV signal will do. Regardless, if it's clipping with a 50mV signal upon entry into the 2nd stage... that's a problem. Right? ...I just want to make sure I'm thinking correctly and any suggestions for how to diagnose or correct would be very much appreciated as always. Thanks in advance.
-
Oh, also note that I experimented with bypassing that gain pot to ground resistor to see if that would result in more clean room. I guess it kinda did, but I didn't see any real benefit to it as far as reducing clipping at the 2 point in the gain pot rotation. So, while it maybe was able to go a bit lower, it still seemed to clip at the same point with our without that resistor bypassed. Just wanted to note that for what it's worth.
-
Assuming a bypassed cathode, as the Vpeak (ie Vp-p/2) signal level at the grid is raised from 0 and gets close to the cathode voltage, the positive going wave will start to clip.
So if the cathode is biased to 2V, the max unclipped signal at the grid will be about 4Vp-p, 2Vp, 1.4Vrms.
With an unbypassed cathode, it will be about twice that, and the clipping will be more rounded.
Partially bypassed will move between the above behaviours as frequency is altered.
I’m sure Merlin covers the 2 clipping mechanisms- grid clipping / conduction and anode current cut off, in the preamp book?
-
Assuming a bypassed cathode, as the Vpeak (ie Vp-p/2) signal level at the grid is raised from 0 and gets close to the cathode voltage, the positive going wave will start to clip.
So if the cathode is biased to 2V, the max unclipped signal at the grid will be about 4Vp-p, 2Vp, 1.4Vrms.
With an unbypassed cathode, it will be about twice that, and the clipping will be more rounded.
Partially bypassed will move between the above behaviours as frequency is altered.
I’m sure Merlin covers the 2 clipping mechanisms- grid clipping / conduction and anode current cut off, in the preamp book?
Well... yes, Merlin does covers a lot of things in his book... doesn't mean I completely understand or remember all of them. ;) Because of these issues and your help, I am forming (albeit slowly) an elementary understanding of how the signal is manipulated by the various stages and components.
For example, I'm now seeing that despite "leaving" the first stage, and prior to "entering" the second stage (incorrect way to conceptualize it I now realize) the signal is affected AT the grid. To demonstrate that for myself, I connected to the signal just after the capacitor after the first plate on V1 and then still at the grid of stage 2 of V1. The signals are markedly different. Even though the signal is "between" stages. As stupid as that may be, it helps me understand that the clipping I'm seeing is due to the Second stage and not the first. The signal after the first plate (i.e. "coming out of the first stage") is not clipped (at least not with the lousy 50mV signal input I have right now). The first pict attached shows a comparison of the two signals at the point where the stage 2 grid signal (the one on top, probe 1) begins to clip on the top half of the wave.
Interestingly, when the gain pot is maxed (and probes set to x10) the stage 1 plate signal looks clipped at the bottom half of the wave, but the stage 2 grid signal is still clipped only at the top half of the wave.
Thoughts on this?
-
Depending on what observations may be gleaned from my last post and the discussion so far, does any part the behavior of V1 look "off" or account for the "nothin but low volume mush" that I'm getting at the speaker? If V1 seems to be behaving as Merlin intended or at least within reasonable expectations, perhaps I should move on to V2? I need some specific ideas of where to look or what diagnostic procedure to perform next.
-
...20V with gain at max, this is with a 50mV input.....
That stage's gain is 50 or 55. So 50mV in makes 2.7V out. So something is wrong with your measurements.
Are there any other DC voltages in there? B+ supply? Cathode? (Grid??) 140V on plate is not wrong but without more data we can't be sure it is biased right.
-
A sidebar for a few things about your scope...
First, which trace is CH1? That's always important for us to know.
Second, Alt/Chop appears to be depressed (in Chop mode). Chop is only useful when viewing very slow waveforms. About the only time you would use Chop mode in a guitar amp would be if you are looking a slow tremolo signals. Need to leave that in Alt mode for most viewing. (This button is disabled when viewing a single channel.)
Third, CH2 INV appears to be depressed. This inverts the channel 2 signal. Mostly should be left undepressed unless you have a specific reason to invert. I suspect you don't.
Fourth, TRIG. ALT appears to be depressed. This causes the scope to be triggered on CH1 for one sweep and CH2 on the next sweep, and so on and on. This happens even though you have selected CH1 as your source for triggering. This mode is rarely useful in guitar or audio amps. Leave it undepressed and then CH1 will supply all the triggers even when CH2 is being displayed. This allows you to see the phase inversion of CH2 in reference to CH1 display. For example, if triggering on CH1 with CH1 probe connected to the input jack, and CH2 is connected to the plate of V1, you will see that the plate is 180° inverted from the signal on the input jack. Viewing phase shifts as you trace through an amp can be very useful. This is a good reason to not use the CH2 INV button.
Bottom line, ALT/CHOP, CH2 INV, and TRIG ALT should be left in the undepressed position unless you know what they do and have a particular reason to use them. CH2 INV can be useful occasionally. One use would be viewing the two out of phase signals from the phase inverter. You can invert the CH2 display and overlay CH1 and CH2 traces to see how closely matched they are. Always keep in mind though, that you are only inverting the display, not the actual signal.
-
...20V with gain at max, this is with a 50mV input.....
That stage's gain is 50 or 55. So 50mV in makes 2.7V out. So something is wrong with your measurements.
Are there any other DC voltages in there? B+ supply? Cathode? (Grid??) 140V on plate is not wrong but without more data we can't be sure it is biased right.
I got a new signal generator (albeit a cheap crappy one), so I believe I got it to be putting out a 100mV sine wave signal at 1000Hz (was a bit squirly at 500Hz for some reason). I checked it with the scope. So with the usual set-up (although with the various buttons un-pushed per Mr. S. Lukey's help) and the Probe 1 (at x1) and Voltage/div set at 5V per square, I'm seeing a 5V output at the V1 Plate 1 (but past the cap so there's no DC on it)... so I'm thinking I've got it set up right and I'm seeing/calculating things correctly now. (i.e. stage 1 appears to be providing a gain of x50) ...or am I?
Did you want me to also take the additional voltage readings at those various points and report them?
-
A sidebar for a few things about your scope...
First, which trace is CH1? That's always important for us to know.
First, thank you for your patience above and beyond the call of duty. I do technical support of various kinds in my line of work (disaster restoration and environmental remediation) and sometimes I think to myself, "where do I even start with this guy?" I assure you that I wasn't nor am not completely high while working with my scope although in retrospect, the evidence does not support that. Again, my thanks for the remedial scope training... I obviously need it. I do think I was just pushing buttons until the scope made nice pretty traces, but didn't change them back... I see now that they were all jacked up.
So... CH1 is the top trace.
Bottom line, ALT/CHOP, CH2 INV, and TRIG ALT should be left in the undepressed position unless you know what they do and have a particular reason to use them. CH2 INV can be useful occasionally. One use would be viewing the two out of phase signals from the phase inverter. You can invert the CH2 display and overlay CH1 and CH2 traces to see how closely matched they are. Always keep in mind though, that you are only inverting the display, not the actual signal.
Got it, I think all the settings are correct now (check in photos below).
So, with things on the scope set up differently/correctly: CH1 is the top trace, set to x1 on the probe, Voltage/Div set to 5V, and connected to Plate 1 of V1 (but after the cap)... so basically just showing the stable signal flowing out of the first stage. CH2 is the bottom trace, set to x10 on the probe, Voltage/Div set to 2 (so every square = 20V if I'm understanding how this all works); it is connected to Plate 2 of V1 (but after the cap)... so showing the signal after stage 2 and before the EQ stuff. The 3 picts below are all set up this way and:
Pict 1 is the resulting traces with the gain set at 25%
Pict 2 is with the gain at 50%
Pict 3 is with the gain at 75%
So... It seems that V1 is just starting to send a clipped signal on to the next stage when the gain pot is turned to around 70%. Doesn't all this seem to indicate that the first two stages (i.e. V1) is functioning within reason? I haven't changed any components from the schematic, but I DO still have V2 bypassed completely. Also I have that resistor to ground on the gain pot bypassed straight to ground, FYI.
-
so I believe I got it to be putting out a 100mV sine wave signal at 1000Hz (was a bit squirly at 500Hz for some reason). I checked it with the scope.
Show me a pic of your scope display with your probe connected to the input jack. Also want to know the VOLTS/DIV setting and whether 1X or 10X probe.
-
so I believe I got it to be putting out a 100mV sine wave signal at 1000Hz (was a bit squirly at 500Hz for some reason). I checked it with the scope.
Show me a pic of your scope display with your probe connected to the input jack. Also want to know the VOLTS/DIV setting and whether 1X or 10X probe.
Got it... Pict 1 attached is of the signal generator probe (red) and the Scope probe (grey) attached to the tip terminal on the input jack (there's a dummy jack inserted so the jack doesn't go to ground). Pict 2 is of the scope. Probe is on x1 for sure and the Voltage/Div is on 50mV, so my read is that 2 squares should be 100mV. Amp is not turned on.
-
Pict 2 is of the scope. Probe is on x1 for sure and the Voltage/Div is on 50mV, so my read is that 2 squares should be 100mV.
Except that your scope is displaying almost 4 divisions! That's 200mVpp. That's twice as much signal as you think!!! This is why I've been questioning your scope skills.
-
Pict 2 is of the scope. Probe is on x1 for sure and the Voltage/Div is on 50mV, so my read is that 2 squares should be 100mV.
Except that your scope is displaying almost 4 divisions! That's 200mVpp. That's twice as much signal as you think!!! This is why I've been questioning your scope skills.
No problem questioning my scope skills... since I have none. So to calculate the voltage one counts from peak to peak, not from the baseline...? So the entire voltage swing is what is considered "signal strength" then, ...correct?
-
The entire voltage swing from one peak to the other peak is known as peak to peak voltage (Vpp)
-
Got it. So, pict 1 (below) is of a 100mV signal then. Probe@ x1, Voltage/Div at 50mV and a swing of two squares. Correct?
Pict 2 is of that 100mV signal with Probe 1 (set at x1) connected again at V1 Plate1 after the cap (Volt/Div set at 5V) and Probe 2 (set at x10) on V1 Plate 2 after the cap (Volt/Div set at 2V). Gain turned up just until the beginning of clipping. About 50%.
Thoughts?
-
.