MOSFET B+ reducer

[R.G.]

I haven't read any of Blencowe's books, but what bits of his advice I've seen in forums seems to be generally OK.

It's important to deduce the objective of the circuit fragments, though. I believe that the circuit fragment as shown is not well suited to your application, dropping 40-80 or so.
That schematic is intended to produce a single zener with a voltage rating of 300V and a high current capability. I believe that one aim of the stacked setup is to divide the power dissipation in a high voltage application, even though MOSFETs able to withstand the 300V in this application exist and are cheap. The series resistor is not optional, as it is what does all the voltage dropping in the circuit. The zener ensures that the full current of the power supply flows through either the circuit being powered, or through the zener itself, so a power zener of any kind makes a LOT of heat.

Your application would always have 1/6-ish the power to be dissipated compared to the 300V amplified zener. It's a different problem.

It's always easier to critique than to design, but to my thinking, it's also oversimplified. In any linear application of MOSFETs it's prudent to put in a series gate stopper resistor of 100-1K ohms to ensure that there are no UHF radio oscillations that most oscilloscopes can't even see, and also gate clamping zeners and diodes to ensure you don't blow through the gate insulation on a transient. I'm sure that Blencowe covered MOSFET best practices somewhere in that book.

Stacked zeners could probably be made to work as a CT B+ dropper, but in a CT zener, the working voltage is in the 40v to maybe as much as 80V range. A single MOSFET is just fine for the application. You don't need high voltage stacking or the constant high power dissipation.

[craftyjam]

I haven't read any of Blencowe's books, but what bits of his advice I've seen in forums seems to be generally OK.

It's important to deduce the objective of the circuit fragments, though. I believe that the circuit fragment as shown is not well suited to your application, dropping 40-80 or so.
That schematic is intended to produce a single zener with a voltage rating of 300V and a high current capability. I believe that one aim of the stacked setup is to divide the power dissipation in a high voltage application, even though MOSFETs able to withstand the 300V in this application exist and are cheap. The series resistor is not optional, as it is what does all the voltage dropping in the circuit. The zener ensures that the full current of the power supply flows through either the circuit being powered, or through the zener itself, so a power zener of any kind makes a LOT of heat.

Your application would always have 1/6-ish the power to be dissipated compared to the 300V amplified zener. It's a different problem.

It's always easier to critique than to design, but to my thinking, it's also oversimplified. In any linear application of MOSFETs it's prudent to put in a series gate stopper resistor of 100-1K ohms to ensure that there are no UHF radio oscillations that most oscilloscopes can't even see, and also gate clamping zeners and diodes to ensure you don't blow through the gate insulation on a transient. I'm sure that Blencowe covered MOSFET best practices somewhere in that book.

Stacked zeners could probably be made to work as a CT B+ dropper, but in a CT zener, the working voltage is in the 40v to maybe as much as 80V range. A single MOSFET is just fine for the application. You don't need high voltage stacking or the constant high power dissipation.

Yeah I though there was maybe an issue with the voltage rating of the mosfet I was using, causing it to fail prematurely. It seems now that the main issue is propper air flow through the chassis. I really would like to avoid instaling a fan, but the b+ dropper is failing with a 200ma load and 50v drop. The mosfet I'm using is an irfp460 which has a significantly higher power rating than the irf820. I have it insulated and bolted to the aluminum chassis with thermal paste inbetween. Even with the chassis upturned exposed to free air, It will still fail. I'm also preventing it from experiencing the large current spikes from the rectifier by taking the CT tap from the negative side of the reservoir cap, as discussed in previous posts. The last thing I have yet to try is adding another mosfet in parallel.

[dorrisant]

Have you tried any other MOSFETs?
The two suggested in the doc from the 1st post work very well.

STW20NK50Z 20A 500v 190w
FQP9N90C 8A 900V 205W

I get it that what you are using might have very similar ratings two the above examples. I do recall testing others and had failures. I pretty much use the STW20NK50Z. I have installed maybe 40 or so. Haven't had a failure yet.

[R.G.]

Yeah I though there was maybe an issue with the voltage rating of the mosfet I was using, causing it to fail prematurely. It seems now that the main issue is propper air flow through the chassis. I really would like to avoid instaling a fan, but the b+ dropper is failing with a 200ma load and 50v drop. The mosfet I'm using is an irfp460 which has a significantly higher power rating than the irf820. I have it insulated and bolted to the aluminum chassis with thermal paste inbetween. Even with the chassis upturned exposed to free air, It will still fail. I'm also preventing it from experiencing the large current spikes from the rectifier by taking the CT tap from the negative side of the reservoir cap, as discussed in previous posts. The last thing I have yet to try is adding another mosfet in parallel.

Hmmm. A 200ma load and 50V across the MOSFET and no rectifier pulses is only ~10W. I looked at the IRFP460 datasheet from Vishay, and it indicates that the IRFP460 is well within the safe operating area for the device. In fact they claim that it's good for amperes at that voltage. So I don't think the current or the voltage rating is at issue.

That leaves temperature, which you're all over, or gate puncturing. Do you have a gate protection zener on it?

Edit:
The 460 also has a thermal resistance of 0.24 C/W with just grease. Grease and an insulator typically adds about 0.5 to 0.7 C/W, so we could guess at a thermal resistance of 0.84C/W to the chassis. That means that for 10W dissipation, the temperature rise from the chassis to the chip inside would be about 10W * 0.84 C/W, or about 8.4C.

We don't know the chassis-to-air resistance. If it were a steel chassis, this could be a problem, because steel isn't nearly as good a thermal conductor as aluminum. The datasheet says that the FET dies at 150C always, but the temperature versus current curve indicates that for under 1A, it's good up to about 140C. With a conservative 40C ambient, you get to an allowable rise of 100C, so the total thermal resistance from chip to air has to be under 10C/W. The chip-to-chassis is on the order of 0.84C/W, so if you can get the final chassis to air resistance down to 9 C/W or less, you should be good. 9 C/W isn't difficult for a large sheet of aluminum like the chassis.

This thumbnailing says that gate puncture is a very likely cause.

Edit 2:
I did some quick estimation of the thermal resistance of the chassis. It comes out to about 3.5 C/W for a 1/8"/3mm thick aluminum chassis. If the insulation and mounting to the chassis is good, heat removal is even more likely to not be the problem. Gate damage is an even higher likelihood.

[craftyjam]

Yeah I though there was maybe an issue with the voltage rating of the mosfet I was using, causing it to fail prematurely. It seems now that the main issue is propper air flow through the chassis. I really would like to avoid instaling a fan, but the b+ dropper is failing with a 200ma load and 50v drop. The mosfet I'm using is an irfp460 which has a significantly higher power rating than the irf820. I have it insulated and bolted to the aluminum chassis with thermal paste inbetween. Even with the chassis upturned exposed to free air, It will still fail. I'm also preventing it from experiencing the large current spikes from the rectifier by taking the CT tap from the negative side of the reservoir cap, as discussed in previous posts. The last thing I have yet to try is adding another mosfet in parallel.

Hmmm. A 200ma load and 50V across the MOSFET and no rectifier pulses is only ~10W. I looked at the IRFP460 datasheet from Vishay, and it indicates that the IRFP460 is well within the safe operating area for the device. In fact they claim that it's good for amperes at that voltage. So I don't think the current or the voltage rating is at issue.

That leaves temperature, which you're all over, or gate puncturing. Do you have a gate protection zener on it?

Edit:
The 460 also has a thermal resistance of 0.24 C/W with just grease. Grease and an insulator typically adds about 0.5 to 0.7 C/W, so we could guess at a thermal resistance of 0.84C/W to the chassis. That means that for 10W dissipation, the temperature rise from the chassis to the chip inside would be about 10W * 0.84 C/W, or about 8.4C.

We don't know the chassis-to-air resistance. If it were a steel chassis, this could be a problem, because steel isn't nearly as good a thermal conductor as aluminum. The datasheet says that the FET dies at 150C always, but the temperature versus current curve indicates that for under 1A, it's good up to about 140C. With a conservative 40C ambient, you get to an allowable rise of 100C, so the total thermal resistance from chip to air has to be under 10C/W. The chip-to-chassis is on the order of 0.84C/W, so if you can get the final chassis to air resistance down to 9 C/W or less, you should be good. 9 C/W isn't difficult for a large sheet of aluminum like the chassis.

This thumbnailing says that gate puncture is a very likely cause.

Edit 2:
I did some quick estimation of the thermal resistance of the chassis. It comes out to about 3.5 C/W for a 1/8"/3mm thick aluminum chassis. If the insulation and mounting to the chassis is good, heat removal is even more likely to not be the problem. Gate damage is an even higher likelihood.

I am using all typical mosfet protections. I think I'm going to go ahead and use a parallel mosfet configuration and add additional vent holes in the chassis. I think the issue is that my build is quite compact with 2 el34's and a gz34 near the chip causing a significant temperature rise. The chassis im using is 17"x6"x2" 1mm thick alluminum. There simply isn't enough surface area for all the heat of 3 big bottle tubes and 5 others.

[R.G.]

I am using all typical mosfet protections. I think I'm going to go ahead and use a parallel mosfet configuration and add additional vent holes in the chassis. I think the issue is that my build is quite compact with 2 el34's and a gz34 near the chip causing a significant temperature rise. The chassis im using is 17"x6"x2" 1mm thick alluminum. There simply isn't enough surface area for all the heat of 3 big bottle tubes and 5 others.

If you have gate clamping protections (... like a zener from gate to source), then you're not having that issue.
It is possible that your chassis isn't enough heat sink then. 1mm/0.04" chassis metal is going to be a considerably worse heat sink, so that looks more plausible. If it were me doing this, I would put a cooking thermometer on the chassis right over the MOSFET and actually measure the chassis temperature to get a better idea what I was dealing with. I ... love... my instant read electronic cooking thermometer for things like this. I equally love my point-and-shoot IR thermometer for most apps, but it doesn't do all that well on uncoated metals.

I just had a stray thought - where did you get your MOSFETs? If from ebay or amazon, there are a lot of counterfeits there. If from Mouser, Digikey, or other industry supplier, they're not likely at all to be counterfeits.

The fundamental reason I keep veering away from paralleling as a solution is that even one of a real, full spec IRFP460 is capable of many times the current involved in this setup. It's possible that there is some other weird thing going on in the circuit to pull big surges, but the CT RMS current in a PT will be less than twice the DC average current, and that ought to be under about 250ma. That 250ma is the max current for a GZ34 in a similar setup. A full function MOSFET of your type is capable of many times that, even including the thermal safe operating area, per the datasheet.

In any case, I hope this works well for you.

[Stephen1966]

I would like to nominate this for the FAQ section.

A lot of expert knowledge on display here and I'm sure it's a common enough problem to qualify.

What say you?

[pompeiisneaks]

I would like to nominate this for the FAQ section.

A lot of expert knowledge on display here and I'm sure it's a common enough problem to qualify.

What say you?

I am fine with it, xtian do you care if I move the thread to faq?

~Phil

[Stephen1966]

Here is some further reading on the same subject I came across. Can't remember if these were linked in these pages already, but here goes:

VVR Install Guidance in 5E3 Tweed Deluxe
https://ampgarage.com/forum/viewtopic.php?f=6&t=19284

VVR noise
https://ampgarage.com/forum/viewtopic.p ... 31#p426531

And an honorable mention from el34world,

VVR and cathode biased amps
https://el34world.com/Forum/index.php?P ... 6#msg60646

[dorrisant]

Does anyone have the schematic for the fixed bias circuit?

[Stephen1966]

Does anyone have the schematic for the fixed bias circuit?

Maybe check out this link:

https://el34world.com/Forum/index.php?t ... 8#msg75788

Search for 'VRM Basic FB.pdf'

[sluckey]

Does anyone have the schematic for the fixed bias circuit?

This may help...

[katopan]

Does anyone have the schematic for the fixed bias circuit?

https://ampgarage.com/forum/viewtopic.p ... vr#p422936

https://ampgarage.com/forum/viewtopic.p ... vr#p418089

[dorrisant]

Does anyone have the schematic for the fixed bias circuit?

Maybe check out this link:

https://el34world.com/Forum/index.php?t ... 8#msg75788

Search for 'VRM Basic FB.pdf'

This is what I was looking for... Thanks!!

[pjd3]

Man o man, this is a gem of a thread, thank you all for your useful and generous input. Amazing.

I'm finishing off a rebuild of my first tube amp which was a Blackvibe 6L6. Nice amp, worked flawlessly except it was way to clean for the live music I play and I sought to get it closer to a typical Fender Blackface vibrato channel with spring reverb (hopefully sound more like a higher powered Deluxe Reverb). So, I'm using various "Single channel Deluxe reverb with no vibrato" builds as guides.

It has a Mojotone Bassman PT which has HT voltage of about 360-0-360 which has put around 460vdc or so on the plates of some Sylvania 6L6WGB's. According to Mike at KCA NOS, I was warned to not exceed 470vdc on these tube but, I figure I can likely get alot more time out of them if I take 20 or so volts off the B+. I just make a B+ Reducer board with 20V Zener to the schematic of R.G's Mosfet follies circuit and still amp trying to figure out the best way to configure it in the chassis. That's not totally clear to me yet even after a good amount of reading and trying to understand the nature of ground possibilities, drawbacks and heatsinking (using chassis as heatsink vs isolating Mosfet and wiring directly to HT centertap/power star ground post). So, trying to see through the myriad of opinions a options to understand which will work best with my amp layout as it is.

Thank everyone, still lots to read here for me.

Best,
Phil D.