SE KT88 High gain amp in UL mode question

[goldstache]

I've built up a SE KT88 4 stage High Gain amp and the OT has a UL tap.  When I measured the voltage at the screen I'm getting 1 volt higher than the plates?  What gives?  I have a 100 ohm 1W resistor in series with the UL tap and I'm getting zero voltage drop across the resistor despite it not looking smoked?

I have an edcor 25W OT with a 3K primary and a 40% screen tap 

My B+ is around 485V

Anybody wanna throw some UL knowledge my way?  The amp sounds great and is a bit overbiased.  Im getting about 20W plate dissipation across the KT.  485 X .045ma.  I know thats above spec but its as low as my adjustable fixed bias pot will go without modifying the bias supply.  Which I plan to do before its all said and done.  I don't want more than 80ma total current draw for the amp and I went to check what current the screens were pulling and was surprised by the voltage on it and the series resistor not having a drop.  This is at Idle.  At full output my plate current creeps up to around 57ma or so. 

Thanks.

[sluckey]

Screen should be higher than the plate just because of the resistance of the primary winding. The primary winding starts at the B+ lead, makes a bunch of turns and arrives at the screen tap. Then the winding continues on with a bunch, bunch, bunch more turns and finally ends at the plate lead. There is a lot more resistance in the wire between the B+ lead and plate lead than there is between the B+ lead and the screen tap.

That resistor should not look smoked. Zero dc volts across that resistor means no dc current flowing through that resistor.

Im getting about 20W plate dissipation across the KT.  485 X .045ma.  I know thats above spec

That tube is capable of 42 watts plate dissipation.

[goldstache]

Thanks Sluckey.  I would assume that there would be current flowing at the screen even at idle.  I must be wrong. 
Somewhere I read that a KT88 in SE is only rated at 19W.  I can't seem to find my source, but I swear I read it somewhere. 

I'll keep probing and digging. 

Thanks for the info!

If you, or anyone else has any thoughts please share.  It's how I learn stuff.

[Paul1453]

>Somewhere I read that a KT88 in SE is only rated at 19W.  I can't seem to find my source, but I swear I read it somewhere. 

This DS confirms your SE Class A 19W output spec.   

https://frank.pocnet.net/sheets/164/k/KT88spec.pdf

This one has much more info on PP UL OT operations.

https://frank.pocnet.net/sheets/084/k/KT88_GEC.pdf

[sluckey]

That data sheet clearly says max plate dissipation is 42 watts. And that's what goldstache referred to.

That 19 watt figure is for typical power output and applies to the other "typical" values listed in that table. Goldstache's figures are quite different.

[PRR]

> UL tap.  When I measured the voltage at the screen I'm getting 1 volt higher than the plates?

Read all three leads.

Say B+ = 500V, G2 = 499V, P = 498V. Isn't it reasonable for the G2 lead to be about half-way between B+ and P?

> I have a 100 ohm 1W resistor in series with the UL tap and I'm getting zero voltage drop

KT88 is an align-grid pentode. G2 current "can" be quite low. Datasheet suggests that 150mA of plate current goes with 3(approx) mA of G2 current.

You are running less that 150mA in Plate. Evenso, assume "3mA". 3mA in 100 Ohms is 0.3V. Your meter may not be able to tell, say, 499.0V from 499.3V. The usual 200.0mV 3-1/2 digit DMM won't. (At >200V, you must be on a "1999.V" scale, no part-Volts.)

> 485 X .045ma

That is WAY cold.

It's half what the tube can handle. You cudda just used a 6L6GC, cheaper.

Divide B+ by load. That is a good first-guess operating point for audio pentodes. 485V/3,000r is 0.162A or 162mA. However, this is 80 Watts Pdiss, far too much.

For 3K load and say 40W Pdiss, you want to wind up near 340V plate-cathode, 110mA plate (or cathode) current. You will about need 35V of grid bias from a cathode resistor, so you want 375V clean B+ at 120mA total (allowing a few mA for little tubes). Your "485V" will surely sag at double current, but you are likely to need an additional R-C filter (always good in SE stage amps) to get the clean B+ down to the 350V-425V range.

> the bias supply

Use Cathode Resistor Bias Whenever Possible!!

An SE stage won't leave class A, cathode current near-constant, resistor is good bias.

Self-bias WILL reduce tube-to-tube variations and make bias adjustments unnecessary.

In this case you have too much raw B+ and can't object to throwing some into a cathode resistor.

[Paul1453]

With your screens not pulling any current and the correct voltage read at that pin.

Somehow it seems your 88 is operating in triode mode as described in those DS to me.   

I'm sure sluckey has much better insight into what is going on with your circuit.
With PRR chiming in, you've got 2 experts guiding you.

I just tried to post a link to info I thought you might be looking for.   

[sluckey]

Neither of those links work.

[Paul1453]

Sometimes that site fails to load on the 1st attempt.

If you just hit the refresh button then it usually loads.

It is working for me on my Army network, so it should be accessible to you guys too.   

The 2nd link is quite large almost 10M.

[sluckey]

They don't work. Click on one of them in the above post.

[Paul1453]

I did and they are working for me.   

Even the 9.32M 2nd link.

If you get something like Page cannot be displayed.

Then click the 2 little arrows button next to the URL, and then it usually comes up.   

[labb]

I built a SE amp that could use a KT88. It can give you about 20 watt output. SE's can be biased at about 90% of max plate dissipation which would be 37.8 watts. With your plate voltage you should look for abut 75 ma across the cathode resistor. With the right speaker it is really loud. Did the same for 6L6's and EL34's. Never cared much for them in a SE amp. Always ended up going back to either a 6V6 or an El84. But then that is just me.

[PRR]

> Neither of those links work.

Work for me.

Maybe your ISP censors Frank's?

[sluckey]

I'm surprised that the Army allows you to surf the internet over their secure network on government time. I know FAA frowns on that big time unless it's work related.

[sluckey]

It's the https protocol that's throwing my browser. I have a workaround.

[Paul1453]

I'm surprised that the Army allows you to surf the internet over their secure network on government time. I know FAA frowns on that big time unless it's work related.

I'm on the NIPR non-secure network here.  The secure network has no connections to the outside world.

As an IT Specialist this type of electronics training site can almost reasonably be justified as work related training.  LOL
I have learned a lot watching you guys help others with complex electronics theory/design and application.   

Much better than my subordinates watching movies on Netflicks on their connections.   

[sluckey]

 

[uki]

They don't work. Click on one of them in the above post.

Works for me, maybe some of us are not able to open the links, I remember having some problems with frank.pocnet so I have been using a different mirror since then.
Try this links it is the same files in a mirror site:
https://tubedata.altanatubes.com.br/sheets/084/k/KT88_GEC.pdf

https://tubedata.altanatubes.com.br/sheets/164/k/KT88spec.pdf

The other mirrors are in this link: http://www.tubedata.info/

[goldstache]

Thanks everyone.  Lots to crunch on there. 
My build has an old Chicago iron transformer (PT).  It's rated 350-0-350 @125ma.  I tried a BJT emitter follower right after my reservoir.  Capacitor multiplier is what Merlin called it.  I used a BUL216 as the semiconductor.  It's very quiet in the power section.  I'm not using a standby switch.
Looking at the data sheet I can't discern how to grasp max current through device in this configuration.  Iceo?????
http://www.mouser.com/ds/2/389/bul216-954433.pdf

So beyond my PT's Max current handling and this device I'm not sure what bias scenario would be best.  I'm not opposed to cathode bias, my fixed bias was just a learning springboard to design and tweak the bias supply.  Though PRR post has me thinking I'm all wrong. 
So to measure RMS I should use my DMM across an 8 ohm load resistor, set up signal gen for 500mv sine.  Adjust amp till waveform starts to clip.  VAC X VAC/8 for wattage of amp????

Also, with no standby, I noticed some cracking when dropping the voltage on my variac to power down.  What should I look out for with this topology?

>reservoir>cap multiplier, cap(OT primary)>R>C>R>C.
If I add a another RC filter to drop B+ I buy some bias range, but might run out of current handling from cap multiplier and PT. 

I may need to hook up a 10ohm resistor after the the rectifier and measure across it to determine total current draw of amp to get a better understanding of its current demands and go from there.  Oh, and the KT's curves need a look too,huh?!?!?
Thanks for getting my wheels turning.  I'd like to maximize output which would be close to the 25W rating of my OT. 

[shooter]

how to grasp max current through device

Ic will keep you outta trouble, but you also need to keep Ptot in mind.  The closer you get to max, the faster you make heat, which lowers, your max, which snow-balls you to smoke.   I like to stay below 70% on most #'s in the sand devices

[PRR]

> Capacitor multiplier

I think you are over-complicating.

> how to grasp max current through device in this configuration.  Iceo?????

Doesn't make sense; that huge package for 0.250mA. Iceo is kinda "leakage", the smallest current you may reasonably control.

MAX current is page 1, Ic, 4 Amps, 4,000mA.

BUT if the voltage across the transistor is not-small, you also have to consider Power and Second Breakdown. Power dissipation on *infinite* heatsink is 90W, so over 22V at 4A would cook it. (And infinite heatsinks are never in stock.) However a 0.125A winding won't deliver 4A. More like say 1.5A for short time. At turn-on, the rectifier pops up to 490V, the filter cap is "empty", 0V. You could have 490V across with over 1 Amp through the transistor. The SOA plot on page 2 takes some deciphering, but 490V 1A is around the 200uS line. It may take more than 200uS to charge your caps. At these speeds, stray winding loss matters. A full computation would be messy. My gut says your BJT peaks at more stress than your KT88, though the average stress may be mild.

[PRR]

> output which would be close to the 25W rating of my OT.

Sine-wave output in class A is at-most half of device idle dissipation.

In tubes, 40% is more likely.

So a 25W output requires 63 Watts idle heat in the tube.

Taking 40W as max diss for KT88, you'll get maybe 16 Watts sine-shape output.

350VAC 125mA DC PT offers 61.8W of DC available, say 60W after deducting for small tubes.

A 60W Pdiss tube at 40% does give 24 Watts audio output. The first-guess loadline is 475VDC/120mADC or about 4K. You don't have either a 60W tube or a 4K load.

Also there will be voltage losses in filtering, and self-bias is good.

Adjusting to 350V 3K 117mA puts Pdiss at 41W, OK for KT88. 16W-17W clean output, 25W-30W grossly clipped.

Bias in UL at 350V will be like 35V, so we need 385V at plate. Allow another ~~35V for drop in OT primary, we need 420V at OT B+ feed. Assuming 440V-480V of raw B+ we need to lose another 20V-60V, at 0.120A, or 150-450 Ohms. A few hundred ohms between first and second filter cap, 2nd cap as 40uFd, gives 6:1 rejection of 120Hz buzz and increasing rejection of higher-order more-annoying buzz.

As the PT's and OT's true parameters are not known, I'd get a bargain-pack of 100R 5W resistors to mock-up the cathode bias and dropping resistor.

[goldstache]

Bias in UL at 350V will be like 35V, so we need 385V at plate. Allow another ~~35V for drop in OT primary, we need 420V at OT B+ feed. Assuming 440V-480V of raw B+ we need to lose another 20V-60V, at 0.120A, or 150-450 Ohms. A few hundred ohms between first and second filter cap, 2nd cap as 40uFd, gives 6:1 rejection of 120Hz buzz and increasing rejection of higher-order more-annoying buzz.

As the PT's and OT's true parameters are not known, I'd get a bargain-pack of 100R 5W resistors to mock-up the cathode bias and dropping resistor.

PRR those calculations, and the insight into how they are arrived at, is most helpful!  It's gonna take me a while for all that to settle.  Thanks!  I will try what you suggest!  I may have the resistors to tinker with. 

{EDIT -- untangled quote - PRR}

[goldstache]

Tried many combinations and recalculations.  To no avail.  I can't get the amp to settle in as prescribed above.

420V B+
350V Plate
35V across Cathode resistor

I was able to drop the B+ to around 420v with a 1K resistor (25W to be safe) in line.  But that was while still in fixed bias wiring.  Every time I experiment with the cathode resistor (unsoldering Fixed bias network) I bring the variac up and check it out and the voltage across the cathode resistor is always above 35V. 

I know I'm shaving off voltage from bad Power supply design,  but I'm interested in the learning experience at this point.  And of course getting my power section performing well for best signal transfer to the OT and max output. 

Anybody got more thoughts on this or want to take me to school on Power supply design as it pertains to this circuit?

To recap, I have around 485V at the OT primary ( I can always add series resistance to get the plate voltage lower albeit poor design)
It's a SE KT88 power section
The OT primary is 3k

[PRR]

> voltage across the cathode resistor is always above 35V. 

What is "above"? 36V? 280V?

What is your cathode resistor?

[shooter]

350V Plate 35V across Cathode resistor

fwiw, here's the values from my 88 SE cathode biased

plate 330v  cathode 29.5v  R=250
16.3W clean, 23ish W distorted

[goldstache]

I got around 65v across the cathode resistor at 1k.
65/1000=65ma

With a 1k in series with B+ it's pretty close to 400v

So I guess it's biased in.  Sorta

Still fuzzy on how to calculate output z for the KT88 in regards to the OT.
I've never used a 1k for a cathode bias resistor.  Makes me think I'm doing something wrong.
Also, haven't worked out effects of UL. 
Thanks for all the guidance, trying to wrap my brain around this!

[shooter]

got around 65v across the cathode resistor at 1k.
65/1000=65ma  With a 1k in series with B+ it's pretty close to 400v

So you have a 1K B+ dropping R AND a 1k Cathode R?
when you say B+ close to 400V, that's plate V also?

I've done 3 '88 SE with plate =<400 and never had Rk >350ohms to get it *close* to 80-100% plate dis.

Still fuzzy on how to calculate output z for the KT88 in regards to the OT.

I *cheat* and look at graphs.  I already know aprox. plate V based on my PT, find *closest* operating values on tube chart, which shows Z for those conditions.

[SILVERGUN]

Do you still have the fixed bias supply hooked up AND the cathode bias resistor connected?

[goldstache]

Fixed bias is not supplied to grid just cathode resistor. 
I will be in front of it later today and post more figures. 
Thanks for the continued help.  I will also reference the data sheet too!!!!

[shooter]

Fixed bias is not supplied to grid just cathode resistor.

do you have schematic penciled up on the amp?
Are you using a PS tap to *inject* a DC voltage at cathode, or using a resistor to ground to develop the bias?

[PRR]

> how to calculate output z

"Divide B+ by load. That is a good first-guess operating point for audio pentodes."

However, you already HAVE a 3K in your hand. Considering cost of shipping iron, your goal is to find a tube operating point that is OK for a 3K load.

> .045ma {0.045A, 45mA}... at Idle.  At full output ... 57ma

SE current should hardly change from idle to full roar. 10% increase is normal. 26% increase means idle current is too low, or load impedance is too low. Increased distortion and power output is dropping. This is under-biased. 

For SE operation you want V/I of the tube to be similar to load  impedance, *and* for best use of big tube dollar you want Pdiss to be like 99% of rated max.

35W tube, 3K load

Trial/error math:
Say 300V. 35W/300V= 116mA. 300V/116mA= 2.57K, a little low.
Say 350V. 35W/350V= 100mA. 350V/100mA= 3.5K, a little high.

So say 324V, 108mA. Check: 324V*108mA= 35W, 324V/108mA= 3K, just right.

Taking 42W as Ab.Max Pdiss (safe for well-made KT88), we get 355V 118mA.

> My B+ is around 485V

Much higher than is needed. I'd lean to a lower-V PT or a higher-Z OT. But shipping iron...

> 65v across the cathode resistor at 1k.

1K ??  Where did that come from?

"110mA plate (or cathode) current. You will about need 35V of grid bias from a cathode resistor"

35V/110mA is 318 Ohms, not 1K!

This is a dartboard approximation. Also it assumes you can get to about 355V+35V= 390V of B+ at the OT plate lead (400-420V at B+ lead). That's why I had suggested
"a bargain-pack of 100R 5W resistors to mock-up the cathode bias and dropping resistor". I think you will end up near 750 Ohms of B+ drop and 300-350 Ohms of cathode resistor.

[PRR]

.

[shooter]

This is a dartboard approximation

You play darts very well
I *land* just about there every time I start where you did

[goldstache]

Wow.  Thank you.  Makes more sense with every helping hand.
Realized a problem from Silvergun's question.  I removed my fixed bias voltage, but never grid biased the KT88 after that.  My 100K bias mixing  resistor was just floating unattached. 
I brought it to ground and mocked up the following from parts already on the bench.

 I used a 660E dropping resistor, and 360E cathode resistor and yielded:

459V B+ node before dropping resistor
389V OT plate node
V across cathode resistor 34.4V
 
So those specs would want a 4.3K OT primary.?

Then I subbed out for 760E dropping, and 360E cathode

457 B+ node before dropping resistor
370V OT plate node
V across cathode resistor 34V

Wants to see 3.9K Primary. 

I'll keep tweaking but I don't have very many values under 100E in large wattage package to really dial it in. 

With my crappy design of brute force series resistance to knock B+ down, is this acceptable to right the amps bias? 

Thanks again for the help. 



   

[shooter]

Wants to see 3.9K Primary. 

your 3K is probably fine, Max Plate dissipation is your biggest concern, (calc Imax=Vk/Rk, then Vplate-Vcathode*Imax gets you close) as long as that doesn't get much above 100%, I'd plug it in and jam, then re-evaluate.

[kagliostro]

Seems that that is about a Power Section


 


Franco

[goldstache]

After getting it pretty close I jammed it for a while and found that I like the sound more when it's biased improperly. 
Probably has to do with perceivable output volume which is louder when wrong?!?!?

Preamp also sounded better with no extra RC filter.  It's off my voltage targets in every way, but sounds better.  I'll have to post the schematic which I nabbed from another forum.  But I think everything is within dissipation. 

It might be a case of guitar player-itus.
Louder being better.

Where it's biased the OT primary wants to see 6.4k. It has only an 8 ohm secondary.  Should I load it with 4 ohm cab to match primary better?????

[shooter]

Should I load it with 4 ohm cab to match primary better?????

my last build pse 6V6 wants 8, I switched from 4 to 16 without *much* change audibly, the #'s squirreled out some, but no where near melt-down.  Keep dialing it, check max, dial, check max

[Willabe]

I used a 660E dropping resistor, and 360E cathode resistor....

What is a 660E and 360E?

[kagliostro]

What is a 660E and 360E?

May be the keyboard is a QWERTY and resistors are 660R & 360R ?

Ciao

Franco

[PRR]

Cathode resistor lower. Nearer 330r.

Put about 5K across your 360 Ohms. 1/2W is fine.

[goldstache]

Pardon me. 
R not E.  Should probably preview my posts:)

Thanks for the help.