VVR, power-scaling question

[jeffhorrigan]

Hi,
    Long time lurker on this and other building boards so my apologies for just popping up with no intro but it seems a high percentrage of knowledgeable people post here so hopefully someone can offer some suggestions to my problem. I have built/rebuilt a couple amps (18watt TMB, Octal Valve Jr) so still a newbie but not completely clueless at this point. I have added a VVR circuit based on Dana's schematic to both amps with great success. My latest project is a 5E3 to which I also added VVR, this time using Dana's board (version 3 I think) which worked fine for a while (4-6 months) but eventually the mosfet broke down which I suspect is due to heat. I generally keep it 3/4 engaged and when it broke down it just became stuck at that voltage. Replacing the mosfet has allowed the VVR to function properly again but I figure it too will break down after a while. I don't want to add a fan because I think that will be too noisy and I don't feel safe running with the back off the amp.

So this conundrum got me to thinking, since I mostly run the VVR  3/4 engaged (scaling whole amp) why not just add a fixed B+ voltage attenuator that I can engage with a switch to drop the B+ down to about 50v or maybe several switches for a few different voltages (ex. 150v, 50v, 30v.) I was thinking of using power resistors to achieve the drop but haven’t yet worked out whether the amount of voltage I want to dissipate is going to be a problem and/or create a fire hazard. My rationale for resistors over the mosfet is that continuously variable power is overkill for me and the resistors should be more robust over the long term than a sensitive mosfet. My question is has anyone else tried this? Did it work OK? Is this a stupid dangerous idea?

Jeff

[dude]

Hi,
    Long time lurker on this and other building boards so my apologies for just popping up with no intro but it seems a high percentrage of knowledgeable people post here so hopefully someone can offer some suggestions to my problem. I have built/rebuilt a couple amps (18watt TMB, Octal Valve Jr) so still a newbie but not completely clueless at this point. I have added a VVR circuit based on Dana's schematic to both amps with great success. My latest project is a 5E3 to which I also added VVR, this time using Dana's board (version 3 I think) which worked fine for a while (4-6 months) but eventually the mosfet broke down which I suspect is due to heat. I generally keep it 3/4 engaged and when it broke down it just became stuck at that voltage. Replacing the mosfet has allowed the VVR to function properly again but I figure it too will break down after a while. I don't want to add a fan because I think that will be too noisy and I don't feel safe running with the back off the amp.

So this conundrum got me to thinking, since I mostly run the VVR  3/4 engaged (scaling whole amp) why not just add a fixed B+ voltage attenuator that I can engage with a switch to drop the B+ down to about 50v or maybe several switches for a few different voltages (ex. 150v, 50v, 30v.) I was thinking of using power resistors to achieve the drop but haven’t yet worked out whether the amount of voltage I want to dissipate is going to be a problem and/or create a fire hazard. My rationale for resistors over the mosfet is that continuously variable power is overkill for me and the resistors should be more robust over the long term than a sensitive mosfet. My question is has anyone else tried this? Did it work OK? Is this a stupid dangerous idea?

Jeff

I assumed you used the mosfet Dana supplied with his updated bd, NTE-2973. They're very good, $9 at Mouser. I too have had this problem a few times in an 18 watt and 5E3, the answers that I got were if install properly no heat related problems. Yet, I have another 18 watt with this same mosfet for two years that I've had no problems and I run it for hours at a time and it's close to the PT way up in a heat trapped area...go figure. I have a PC fan on it. Although so did the others...? The 5E3 is a steel chassic so maybe that's it..? And the other is a conversion from a cheap thin chassis, probably the reason.

I try to install the mosfet away from the PT heat although hard to do as the hole is close. I've had good success with a PC processor fan and it's heat sink (the small old ones 2" x 2") on the opposite side of the chassis. They fit well are noiseless and keep the mosfet cool. You can power it off the 5v tap if you have one not being used in an 18 watt or off the heater supply (draws hardy nothing) you'll have to rectify it to dc, I have several schematics to get about 5 or 6 vdc. The fan runs a lot slower at that voltage so perfect.

al
 












     

[jeffhorrigan]

thanks!
my 18watt VVR is bolted to a steel chassis and that has lasted so far 2 years. The 5E3 chassis is stainless steel and I wonder if it just doesn't conduct heat very well. I had it bolted to the chassis half way between the PT and the eyelet board. Perhaps a fan is going to be the way to go after looking into the resistor idea a little more it looks unfeasible as I would need 200-300 watt resistors which are going to put out a lot of heat.

[dude]

I think my 5E3 is stainless steel too, the mosfet lasted about a few months. When they go you usually have almost all your B+. I never open-up the amp after it blew. I did have PC fan on it, maybe it stopped working but I doubt it. I'll replace it when I get to it, maybe I'll scrap the chassis next time. 

I would use the PC fan as a safe measure, so many older PC's use that 2" square fan and the heat sink is small too, fits easily. There all over the place basically for free. I tapped my heater supply for the voltage but I remembered the voltage was too low 3 volts and it didn't spin Dana helped me with another way to rectify the tap. I think I got 5 volts and that was perfect, very quite. I saved the diagrams I got from Dana, let me know and I'll post them, just a few IN4007's and a small cap. There are some other cheaper mosfets that work, do a search here.

al

[FYL]

My latest project is a 5E3 to which I also added VVR, this time using Dana's board (version 3 I think) which worked fine for a while (4-6 months) but eventually the mosfet broke down which I suspect is due to heat.

Hmmm. Dana's kit is OK in a SE3, provided that the Mosfet is properly mounted. Thermal problems manifest themselves in a matter of minutes, not months.

I'd rather suspect oscillations leading to self-destruction. Are the leads short and tightly twisted? Is the Mosfet as close as possible to the pot? Do you use a gate stopper?

 

So this conundrum got me to thinking, since I mostly run the VVR  3/4 engaged (scaling whole amp) why not just add a fixed B+ voltage attenuator that I can engage with a switch to drop the B+ down to about 50v or maybe several switches for a few different voltages (ex. 150v, 50v, 30v.)

Have a look at the Geezer switch - split cathode Rs and switchable caps and parallel R. You get three levels of power with full power stage disto at all settings using only a handful of components.

[jeffhorrigan]

thanks for pointing me to the geezer switch, it sounds like what I'm looking for. As for oscillation on the leads to the mosfet I was not aware that could be an issue. As it is DC voltage I did not think twisting would be necessary. However it is true that in the 5E3 the mosfet is mounted some distance from the VVR board and POT (about 4inches) whereas in my 18 watt the mosfet and other components are co-located and the volume POT is some distance away (about 5-6 inches.) I don't have any known oscillation problem in the 5E3 (at least none I can hear) and the VVR board and mosfet are on the "power" side of the chassis away from anything that they could induce noise into. I have the mosfet mounted in the coolest place I could find in the bottom left corner of the chassis when it is installed in the cabinet so that it is not directly above any heat producing component, if the problem is heat related hopefully that will solve it. The mosfet causes the outside of the chassis to get very warm in that spot so I've added an aluminum heat sink to help dissipate the heat from the chassis at that point. The mosfet is now in place where I can easily get at it to swap it out and I'm thinking of adding sockets to the wires so I won't even have to solder to make the replacement. Just apply the heatsink compound and screw the new one in.

[HotBluePlates]

As for oscillation on the leads to the mosfet I was not aware that could be an issue. As it is DC voltage I did not think twisting would be necessary. ...  I don't have any known oscillation problem in the 5E3 (at least none I can hear) ...

You won't hear these oscillations. They are simple too high frequency.

I have a adjustable regulated power supply made by Hewlett-Packard, maybe in the mid-50's. Guess what tubes it uses for the main regulator circuit? 6L6's to pass the current supplied by the regulator, and common preamp tubes (like a 12AX7) in the control circuit.

It does not matter that the circuit is designed to handle d.c., just as it would not matter whether you actually played your guitar if your amp was oscillating. It is a potential problem because the MOSFET may have high gain well in the RF range. When you build, you will do well to always assume oscillation will be a problem, and the the preventive measures against it before you actually see symptoms.

So take the steps FYL suggested. Leads should be as short as possible and twisted, and include a gate stopper for the MOSFET.

[FYL]

Just apply the heatsink compound and screw the new one in.

I guess that you also use a proper thermal pad between the Mosfet and the chassis. If so it should also be changed - except if you use ceramics-based pads. Mica and silicone pads can very easily get broken or punctured, conductive thermal grease can creep, etc. All leading to partial shorts after a while.

[Willabe]

I was under the understanding you should not use thermal grease with a silicon pad?     


            Brad       

[FYL]

Quality loaded silicon pads should be used alone, basic models - ie. generic Chinese crap - and micas require thermal compound or grease.

[Willabe]

I just ordered some from Mouser. Do you think these are good? I realy would rather not have to mess with the thermal grease if I don't have too.

I can't figure out how to post the link to the catologe page, sorry.


                  Wakefield TO-3 Kapton .006" reinforced insulator. Part # 567-175-6-310P.


      Thanks,    Brad     

[FYL]

Wakefield TO-3 Kapton .006" reinforced insulator. Part # 567-175-6-310P.

The 175-6s are high quality greaseless insulators. Good stuff.



From the Wakefield catalog.

[Willabe]

Ah, good. Got something right.     

Thanks FYL.


           Brad       

[jeffhorrigan]

so if the mosfet is about 4 inches of wire away from the volume pot but the wires are twisted and I have a gate stopper resistor will that likely dampen any oscillations sufficiently? Would I be able to see the oscillations on an oscope or as voltage fluctuations with a DMM? Also, I will search the forum but, what size stopper resistor should I use? Something like a 68K as you would use for a grid stopper? A shame the VVR instructions from Dana make no mention of this issue. The implication from the instructions suggest mounting the POT and mosfet in different locations apart from the board is perfectly reasonable. "The pot and mosfet are not installed on the board so that you can install the board, pot and mosfet in different locations within the amp." Although when I received my Kit they were both attached to the board so maybe the instructions are out of date?

Thanks for everyone's helpful suggestions!

[jeffhorrigan]

So in looking at Dana's VVR schematic it looks like there is a 100k gate stopper on the board which if I understand correctly should be moved off the board to the gate pin of the mosfet itself to ward off oscillation (along with twisting the wires.) Does that sound right?

[FYL]

So in looking at Dana's VVR schematic it looks like there is a 100k gate stopper on the board which if I understand correctly should be moved off the board to the gate pin of the mosfet itself to ward off oscillation (along with twisting the wires.) Does that sound right?

Yup. The gate stopper should be as close to the gate pin as possible - I usually solder it to the pin itself.

[FYL]

Would I be able to see the oscillations on an oscope or as voltage fluctuations with a DMM?

Mosfets are wide bandwidth devices, most are optimized for switching applications, with typically Ft > 40 MHz. You need a wide bandwidth scope, say at least a 100 Mhz model. Better yet: a spectrum analyzer...

Also, I will search the forum but, what size stopper resistor should I use?

It all depends on the actual Mosfet used, the circuit, it's topology and the amp. Stoppers are usually 10K to 1M > 1 W/500 V resistors with 100K giving excellent results in most situations.

[Willabe]

Is'nt it important for the 12v zener across the source/gate to be as close as posilble too, for the same reason? (That is if the mosfet does'nt have one already built in?)


      Thanks,    Brad       

[FYL]

The zener can be located on the PCB or soldered to the pins. It only acts as a clamp protecting the Mosfet.

[phsyconoodler]

By the way,a 5E3 chassis is chromed steel(unless it's from Weber) and it holds heat like crazy.A heat sink is suggested and a fan.

 Sometimes we have to think about when this amp was designed and think about how hot a 5E3 gets without any way for the heat to get out.A fan won't add noise if done properly.

[HotBluePlates]

Would I be able to see the oscillations on an oscope or as voltage fluctuations with a DMM?

This is a silly tidbit at this point, but I really like old vacuum tube HP test gear.

You know what two things made Hewlett-Packard the company it is today? [Hint: it weren't computers]

They designed and built oscillators (originally audio and super-sonic ranges) and also built an a.c. only voltmeter which was capable of reading low-RF voltages. I have some old HP oscillators, and if I connect a 50's era ac-only VTVM to the output, as well as a nice Fluke 87III, then keep turning the frequency higher and higher, I will reach a point where the fluke tells me the voltage is falling while the HP VTVM shows that it is staying steady.

Point being that all meters measuring a.c. (or any other test gear) has a limit to its bandwidth, meaning a point at which it no longer can detect the voltage present. I don't remember offhand (though I'm sure it's in the manual), but the Fluke starts giving a drooping response around 100-200kHz. Oscillation may come as bursts of RF waaay above that, so you really need a scope to see what's happening.

Course, not all scopes are created equal. You have to know enough up front to pick the best tool.

Now where did I put my hammer?