Problems with VVR

[Leevi]

I'm installing a VVR to an old VOX AC30 and have met some problems.
The circuit works fine but the temperature of the MOSFET increases quite fast.
The MOSFET burns after some minutes playing.

I have bolted the MOSFET (with isolation) to the chassis.
Could a possible oscillation be reason for this?

I'm regulating the power tubes only and use diodes around the VRR.
Any tips?

/Leevi

[Ed_Chambley]

Pull an old heatsink out of a dead computer power supply if you have one.  Put some thermal grease lightly on both sides of the MOSFET pad and then mount it to the chassis.  The MOSFET is going to get hot.  If you have already mounted a heatsink, then a fan is the next step.  I have tried VVR's in a few amps and while many people love them, I have found I do not care for them.  It does work, but I put one in a 40 watt Super Reverb on a huge heatsink outside the chassis and wired a 4" 7volt fan right next to it.  Still got hot.  I haven't had much luck with VVR's in higher wattage amps,

[Leevi]

Pull an old heatsink out of a dead computer power supply if you have one. Put some thermal grease lightly on both sides of the MOSFET pad and then mount it to the chassis.

Should I isolate the heatsink from the chassis or bolt it to the chassis and isolate the MOSFET from the heatsink?

/Leevi

[tubenit]

Is your chassis aluminum or steel.  My understanding is aluminum works better for a heat sink.

Tubenit

[thelonious]

[dangerously wrong information removed]

[Leevi]

am under the impression - and perhaps I am wrong - that the metal tab on a mosfet is intended to be in direct contact with the heat sink and/or the chassis. I don't see why it would have to be isolated unless the metal tab has some electrical connection to the mosfet. Does it? Any isolation you do will contribute to the heat problem...

It must be isolated since the high voltage is on metal tab

Btw. I have solved the problem.
I was only regulating the plates of the power tubes and not the screens.
That caused the rapid temperature rise. After I changed that the temperature increases but very slowly.

/Leevi

[thelonious]

Glad you solved it!

[Willabe]

I am under the impression - and perhaps I am wrong - that the metal tab on a mosfet is intended to be in direct contact with the heat sink and/or the chassis. I don't see why it would have to be isolated unless the metal tab has some electrical connection to the mosfet. Does it?

Yes it is connected electrically to the inside of the mosfet and has the FULL voltage on it in this case of the B+ power supply.     

It must be isolated from the heat sink and/or chassis.

It's the same with a metal back/tab transistor.


                         
                                 Brad         

[thelonious]

Sorry, guys - showing my ignorance. Thanks for helping me understand. I never realized that the thermal compound in the solid state stuff I've worked on was also an electrical insulator.

[Leevi]

Is your chassis aluminum or steel.  My understanding is aluminum works better for a heat sink.

Yes, aluminum is better. Since the VOX AC30 chassis is steel I added some extra aluminum heatsink around the MOSFET.
I spent lot of time for cleaning the chassis in order to get the thermal contacts better.
/Leevi

[Willabe]

I never realized that the thermal compound in the solid state stuff I've worked on was also an electrical insulator.

No.

You still need to use a mica sheet or a sil-pad as an insulator . I like the sil-pad better because then you don't need the themal grease.

The thermal grease is used to fill in the micoscopic voids in the mosfets tab and the heat sink/chassis. Your not supposed to apply it like mayo on bread or cake frosting. You put on a thin coat and use the side of, say a razor blade to gently scape off the excess so only a very thin coat is left that fills in the voids.

And some brands of themal grease are conductive, which is a bad thing.


                
                                Brad    

[Ed_Chambley]

A good rule of thumb is if you are wondering it it needs isolation, do it.  I always mount MOSFTS with nylon nuts and bolts.  You can use the nylon washers, but I don't trust it.  If it is bolted up with nylon there is no change of the metal nut and screw making contact.  Just mu thought, other use that little metal bolt and bi-level washer.

[Leevi]

I share here the schematic i have used for the VVR.
Only power tubes are regulated
/Leevi

[Fresh_Start]

I share here the schematic i have used for the VVR.
Only power tubes are regulated
/Leevi

Now I'm confused.  That schematic looks like both the power tube plates and screen grids ("screen resistors")  are regulated. 

Respectfully,
Chip

[Leevi]

Now I'm confused.  That schematic looks like both the power tube plates and screen grids ("screen resistors")  are regulated.

Yes, that's the idea. I first "forgot" to regulate the screen (pin 9 of EL84) which caused the problem described in this topic i.e. overheating/burning of the MOSFET.
/Leevi

[Ed_Chambley]

Do you like it better than a PPIMV?

[Leevi]

Yes, I find it better and more usable than PPIMV.
/Leevi

[jon9]

I know this topic is from years ago, but maybe someone will see it and can give me an answer. I am having the same issue with an ac30. I put in vvr, just to the plates (sounds great)and the mosfet gets very,very hot. I have it on a small heat sink and was going to increase the size. But why would the addition of regulating the screens make the mosfet run cooler? They are recieving the lowered voltage, after the mosfet?

[kagliostro]

Give a try, use the VVR only on screens

Franco

[Colas LeGrippa]

Hola, amigos !

I have tried once to install a VVR in one of my amps. I have spent too much time tweaking it( + money ) for a very deceiving result. If the goal of doing that is to reduce the overall db of the amp with the same overdriven tone, I think a power attenuator is a better choice ( Marshall's power brake. THD o others....one other option is a VARIAC , they're cheap now). In my appartment I use a distortion pedal cranked almost to it's maximum , with the amp turned down to audible sound. I don't get the same tone out of my amp as if it was cranked up much louder obviously, but for practicing, it's good. 

Colas LeGrippa

P.S.: keep up the good work and love your woman ( did you kiss her b4 goin' to work this mornin' ? )

[jon9]

Give a try, use the VVR only on screens

Franco

Why would this lower the heat to the vvr mosfet, and would just regulating the screens sound different from just the plates because just the plates gives me almost exactly the sams sound.  I've tried many attenuators over the years and I hate the sound of them all. Ive had luck with the eminence maverick speakers on my '66 bassman. They work great.  The amp I am working with for the vvr is an ac30c2x. Here is the schematic..http://music-electronics-forum.com/attachments/35600d1441677204-ac30c2.zip

[kagliostro]

Lowering only the screen results in lower output power but there is much less stress on the mosfet because of the low current that is on screens respect to the plates

give a look here

https://www.tube-town.net/cms/?DIY/Amp-Tools/VoCo_Variable_Voltage_Control

http://www.tube-town.net/ttstore/Kits/Power-Supply/Kit-TT-VoCoM-Variable-Voltage-Regulator::5651.html


Franco

[jon9]

Thanks! I'll try it. But what about the sound difference? Because then why lower the plate and screen as some do?

[kagliostro]

About sound you must try it and listen, then you decide if you like or not

Franco

[jon9]

Ok, I regulated just the screens and the mosfet barely gets warm, so that's good, I don't have to worry about that. But I don't think it sounds quite as good and it really doesn't change the volume until it's about half voltage on the screens and after that there's only a little usable area. Too much and it doesn't sound that good. As the screen voltage goes down the plate voltage goes up. The range on the plates is 309v with the vvr off to 393v with it at max. But this will have to due for now unless I can find another solution so the mosfet doesn't cook. I just need to knock a little volume off the ac30c2x as it's massively loud and the master volume needs to be at least halfway up or more to bring the good bottom end into it. I just need it to have a useable stage volume.
 

[DummyLoad]

Sorry, guys - showing my ignorance. Thanks for helping me understand. I never realized that the thermal compound in the solid state stuff I've worked on was also an electrical insulator.

it is but you need mica washer and compound for proper thermal management. there are other materials used for insulation washers, however, mica is the most common.


ALSO: if you're mounting a TO-220 tab, you need a nylon shoulder washer to insulate the bolt, or use nylon fasteners.


--pete


 

[2deaf]

Which MOSFET are you using?

Sometimes they go into oscillation.  You could try a 300 ohm gate-stopper resistor soldered directly onto the gate pin with the leads cut as short as possible.

As Kagliostro said, lowering the screen voltage will lower the plate current and the output power will decrease.  Pentode plate current doesn't change very much with different plate voltages when the screen is held constant (until you get to the knee).  So just lowering the plate voltage will still put a strain on the MOSFET. 

You will probably want to adjust the screen and plate voltages at the same time to get the desired effect. 

[Leevi]

Use a thin fiber plate (<1mm) between the chassis and Mosfet.
Add heatsink paste to the contacting layers and bolt the Mosfet properly to the chassis.
This should be enough.
/Leevi

[jon9]

Which MOSFET are you using?

Sometimes they go into oscillation.  You could try a 300 ohm gate-stopper resistor soldered directly onto the gate pin with the leads cut as short as possible.

As Kagliostro said, lowering the screen voltage will lower the plate current and the output power will decrease.  Pentode plate current doesn't change very much with different plate voltages when the screen is held constant (until you get to the knee).  So just lowering the plate voltage will still put a strain on the MOSFET. 


You will probably want to adjust the screen and plate voltages at the same time to get the desired
effect.

I am using the nte 2973. What makes you think its going into oscillation? 
I would like to try regulating the plates and the screens but I am concerned about the mosfet heat as I stated in my first post when I regulated just the plate the mostfet and heatsink temp went through the roof.
Also I'm a guitar guy that likes to mess with my amps to get the best sound. So I don't have a ton of knowledge.  If I want to regulate both would I just combine the 2 B+ legs into and out of the vvr or do I need to add some components also? (I posted scematic in prev post) thanks!

[dude]

Using an old PC's processor fan and it's heat sink (fan is about about 2" square) works well. Use the filament tap, rectified for about 5 or 6 vdc to the fan. This spins the fan slower and can't be heard. Those old heat sink processor fans have a clip to mount the fan to the heat sink or four small screws. I have the schematic for the 5 or 6 vdc tap ask me if you need it..

Use grease on both sides of the mosfet pad, don't let the grease over-flow to the chassis, also grease the back of the heat sink mounted under and opposite the VVR on the outside of the chassis.

Now when you turn the amp standby on the fan comes on keeping the mosfet cooler. Only problem is the standby hole is usually used for the 1m pot of the VVR and this is usually locate around the PT which creates heat but that little fan will keep it cool.

I have a 5E3 with a steel chassis and kept blowing the mosfet till I started using these heat sinks and 2" square fans, they are all over the place in any old PC.

al 

[2deaf]

I am using the nte 2973. What makes you think its going into oscillation? 

Presuming the MOSFET is sufficiently heatsinked, it shouldn't have any problem in your application.  The NTE2973 is rated 900V at 14A and a big boy like that ought to be able to handle a direct short at the Source when powered by your transformer.  Yet you report rapid heat buildup, so I am entertaining the possibility that it is going into something like RF oscillation.

The NTE2973 has a TO-3P package which should have the bolt hole insulated from the drain so that you can use regular steel screws to mount it.  An Aavid 43-77-20G polyimide plastic pad works well with this package.  The chassis alone should be a sufficient heatsink.  If not, dude has a good approach.   

[jon9]

No, there is not a problem. Well....When I only regulated the Plates the mosfet, heatsink and surrounding got very hot. (like Leevi in the original post) then someone suggested that I only regulate the screens which I did. That solved the heat issue but I thought it maybe sounded a bit better and had more control with just the plates. So then it was suggested that I do both, and as I stated I didn't know exactly how to do that if I tie the B+ ins and outs from both legs(b+ to ct of OT to plates and B+ to screens) going into the vvr or do I need to add something else . and would that again generate the heat issue?

[kagliostro]

I would like to know why a lot of people go on using the NTE2973 that has a really high price instead of all other mosfet that has low price

Franco


p.s.: Give a look to the IRFPE50

http://www.digipart.com/part/IRFPE50

the IGBT FGA20S120M (1200v 40A) seems something to give a trythe

the STW11NK100Z (1000v 8.3A) at € 3.80 ($ 4.98) seems a good candidate to substitute the expensive NTE Mosfet

[jon9]

I used it only because i just bought an ac30c2x and it needs to be at least halfway up to get the good full sound but that was going to be way loud for stage volume and I HATE attenuators so I looked into it and saw vvr as a solution bought a kit from Hall and promptly broke a leg off the mosfet and looked into what I should use and it said the NTE one so for 9 bucks I thought it was a bargain compared to a shitty 300$ attenuator? But now I see there are cheaper ones(i have 4 IRFPE50 their way from ebay for like 8 bucks for all 4
 incase I busted up another...

[2deaf]

if I tie the B+ ins and outs from both legs(b+ to ct of OT to plates and B+ to screens) going into the vvr or do I need to add something else

You would branch off after the first filter following the rectifier.  The attachment shows the basic topology with one branch going to the PI and preamp and the other branch going to the plates and screens.  Nevermind the different values, the different pot, the BJT current limiter, and the bias circuit.  Just look at how the plate/screen supply branches then goes through the MOSFET, then through the 1N4007, then to the plates, then through the 1K resistor to the screens and at the placement of the filter caps.

[jon9]

I see, but quite a bit different than mine. I understand and I don't... could you take a look @ my amp schematic below and give the same advice? I like to be very careful and sure.

[2deaf]

I see, but quite a bit different than mine. I understand and I don't... could you take a look @ my amp schematic below and give the same advice? I like to be very careful and sure.

It's the same thing, they just drew it all weird.  Give me a little time and I will hand draw a possible schematic.

[PRR]

> vvr, just to the plates and the mosfet gets very,very hot. I have it on a small heat sink

The MOSFET must waste-off all the power you DON'T want going to the tubes.

Stock, you have four tubes rated 12W each so I would bet they take 48 Watts total.

When you add drop to a resistance divider, worst-case is half voltage. Here the 4-tubes get 12 Watts total and the MOSFET gets 12 Watts.

12 Watts is a LOT for a small MOSFET. It needs a BIG heat-sink. Many chassis are not really enough meat and area to spread-out 12 Watts of heat and keep the MOSFET safe.

By my conservative rule of thumb, 12 Watts suggests 24 square inches, or 12 square inches exposed to cool air both sides. Say 3.5"x3.5" (9cm*9cm). Average heat-flow distance is almost 1 inch, thickness should be near 1/10th of the distance, so 0.1" (2.5mm) metal (more than average chassis stuff).

Modern MOSFETS "can" run a lot hotter than I would like (based on burning-up a lot of 1970s parts). So a smaller sink may work. But you can NOT cheat this too far.

You almost certainly want to drop screens along with plates. For small drop, no harm, but unexpected action. For an extreme drop you could burn-up the screens when the plates stop taking current.

VVR-ing *just* the screens is a thing. As screen current is 10%-20% of the total, it is a lot less heat. But that only reduces tube current, not voltage-swing into hi-Z speaker impedance peaks. Much will depend on the speaker impedance curve.

[jon9]

> vvr, just to the plates and the mosfet gets very,very hot. I have it on a small heat sink

The MOSFET must waste-off all the power you DON'T want going to the tubes.

Stock, you have four tubes rated 12W each so I would bet they take 48 Watts total.

When you add drop to a resistance divider, worst-case is half voltage. Here the 4-tubes get 12 Watts total and the MOSFET gets 12 Watts.

12 Watts is a LOT for a small MOSFET. It needs a BIG heat-sink. Many chassis are not really enough meat and area to spread-out 12 Watts of heat and keep the MOSFET safe.

By my conservative rule of thumb, 12 Watts suggests 24 square inches, or 12 square inches exposed to cool air both sides. Say 3.5"x3.5" (9cm*9cm). Average heat-flow distance is almost 1 inch, thickness should be near 1/10th of the distance, so 0.1" (2.5mm) metal (more than average chassis stuff).

Modern MOSFETS "can" run a lot hotter than I would like (based on burning-up a lot of 1970s parts). So a smaller sink may work. But you can NOT cheat this too far.

You almost certainly want to drop screens along with plates. For small drop, no harm, but unexpected action. For an extreme drop you could burn-up the screens when the plates stop taking current.

VVR-ing *just* the screens is a thing. As screen current is 10%-20% of the total, it is a lot less heat. But that only reduces tube current, not voltage-swing into hi-Z speaker impedance peaks. Much will depend on the speaker impedance curve.

You certainly know more about this than I, But, Lots of people are running these things with out blowing them up And without a heatsink bigger than their amp? I'm just trying Not to blow up my amp.

[2deaf]

My drawing is very poor to start with and it only gets worse when I draw right to left.

[jon9]

The original poster Leevi said that he just did the plates and the mosfet got way hot(like me) Then he said he did the plates And the screens and it didn't get that hot any more. I would like to know Why that is the interaction between the 2 made it less hot, and if it is so I would like to do that. But I'm not sure if I have to do anything other then combine the 2 B+ legs (plates and screens) or if I have to add something in between (diodes, resistors, caps)Below I marked up the schematic. Where I had it(just plates, hot mostfet sounded great) where I have it now(just the screens, Sounds real good, less control cool mosfet&heatsink) and the question of how to do both.. Like I said I'm a guitar guy that likes to do his own amp stuff, I know it's crude I just like to fully understand and do it right and safe.

[jon9]

My drawing is very poor to start with and it only gets worse when I draw right to left.

haha Thanks so much 2deaf. Unfortunately. This is not what I do. This amp with the pcb and the way it is laid out compared to the schematic just doesn't make much sense to me. My '66 bassman, marshalls and my Fargen all make sense totally when I compare to schematic. (because they're not pcb and there is big ass components)For this one I need more of a "layout" like they did on the fenders to get it. Again this is not what I do. I know how to be safe with tube amps I've been playing guitar for 40 years and tubeamp almost just as long(only tube amps)and been messing with and modding as long as i've owned tools. I like to do this stuff myself.

[2deaf]

Seems to me that if you can hack into B+1 (plates) and into B+2 (screens), then you can do it like I showed.  Just leave R120 alone along with the rest of the power supply.

Another thing to consider would be to use two MOSFETs and divide the load between them.  PRR showed a pretty slick way to control two MOSFETs with one pot on some other thread not all that long ago.

[jon9]

Seems to me that if you can hack into B+1 (plates) and into B+2 (screens), then you can do it like I showed.  Just leave R120 alone along with the rest of the power supply.

Another thing to consider would be to use two MOSFETs and divide the load between them.  PRR showed a pretty slick way to control two MOSFETs with one pot on some other thread not all that long ago.

I can, basically the whole amp is pcb but there IS a wire going to plates and a wire going to the screens.  So I can just put them together going in and out of the vvr? Thanks!

[jon9]

My drawing is very poor to start with and it only gets worse when I draw right to left.

Thanks for your help 2deaf. As I said, I want to do this right. In that schematic you drew up you added a few things. If you have the time could you tell me why as I would like to understand how they affect the cicuit.  You added a 330k resistor,  a cap,  you wrote down "100" should I assume its another filter 100uf 450v? Finally you moved the 1k 10w and put a 4.7 in its place. Thx!

[2deaf]

In that schematic you drew up you added a few things. If you have the time could you tell me why as I would like to understand how they affect the cicuit.  You added a 330k resistor,  a cap,  you wrote down "100" should I assume its another filter 100uf 450v? Finally you moved the 1k 10w and put a 4.7 in its place.

The 100uf/450V capacitor at B+2 serves the same function for the screens as C68 did before the VVR was installed.  The 330K resistor is just a bleeder.

R120 (1K 10W) originally had the screen current and the preamp current going through it causing a voltage drop across R120.  I changed R120 to 4.7K to get a larger voltage drop from the preamp current alone in order to compensate for the lack of screen current going through it after the VVR was installed.  You can leave R120 alone since it is in the PCB and a pain to change.  The preamps will be running at a higher voltage, but probably no big deal. 

[jon9]

The 100uf/450V capacitor at B+2 serves the same function for the screens as C68 did before the VVR was installed.  The 330K resistor is just a bleeder.

R120 (1K 10W) originally had the screen current and the preamp current going through it causing a voltage drop across R120.  I changed R120 to 4.7K to get a larger voltage drop from the preamp current alone in order to compensate for the lack of screen current going through it after the VVR was installed.  You can leave R120 alone since it is in the PCB and a pain to change.  The preamps will be running at a higher voltage, but probably no big deal.

Thanks! Leave it alone meaning not replace it w/ the 4.7? Shouldn't be a big dealbig deal to do that. But Should I move it/include it(or get A new one) for the power tube vvr'd supply? or not even put it there?

[2deaf]

Thanks! Leave it alone meaning not replace it w/ the 4.7? Shouldn't be a big dealbig deal to do that. But Should I move it/include it(or get A new one) for the power tube vvr'd supply? or not even put it there?

It sounded to me as though you could leave all of the power supply intact on the PCB and only alter the plate wire and the screen wire.  If so, disconnect the existing screen wire and use the plate wire as I diagrammed.  You will need to get a new 1K/10W resistor, a new 100uf/450V capacitor and a new 330K resistor plus whatever you need for the VVR itself.

[jon9]

Yep, thank you.

[jon9]

Hooked it all up working and sounding good. Thanks!