MOSFET B+ reducer

General discussion area for tube amps.

Moderators: pompeiisneaks, Colossal

Post Reply
User avatar
xtian
Posts: 6990
Joined: Mon Apr 19, 2010 8:15 pm
Location: NorCal
Contact:

MOSFET B+ reducer

Post by xtian »

A big thank you to dorrisant for the PDF and DIYLC docs attached here.


--

This idea is often thrown out as an inexpensive way to reduce B+ voltage, but I can't find any practical, proven examples. So I hope to detail my build here for the community.

I want to implement the MOSFET B+ reducer shown at the bottom of this page written by R.G. Keen. Hey, RG!

I'm using a Hammond 272JX with full wave solid state rectifier, capacitor input load, to power a pair of 12ax7 and a pair of 6L6GC in cathode bias. Thinking I'll see 0.71 x 600 =~ 420v B+.

I'd like to drop 0 (no reduction), 50, and 100v using the MOSFET B+ reducer.

[img:931:401]http://www.geofex.com/Article_Folders/m ... osfol4.gif[/img]

1. How much heat will the MOSFET dissipate? Just guessing: if it drops 100 volts at 200mA, that's 20 watts, which matches RG's estimate in the article. I chose a (hopefully) overspec'd N-channel power MOSFET, 900V, 8A, 205W.

2. How do I choose the Zener? Do I choose a 50v Zener for a 50v drop, and a 100v Zener for a 100v voltage drop?

3. How do I choose the two resistors shown to match the desired voltage drop?

4. Can the MOSFET get bolted directly (without mica insulator) to the chassis to serve as heat sink? In other words, can the back of the MOSFET be grounded?

Thanks in advance!


[EDIT: Updated doc attached with correction by @dorrisant]
MOSFET B+ Reducer.pdf
You do not have the required permissions to view the files attached to this post.
Last edited by xtian on Thu Jul 12, 2018 7:04 pm, edited 3 times in total.
I build and repair tube amps. http://amps.monkeymatic.com
User avatar
lord preset
Posts: 447
Joined: Tue Aug 27, 2013 3:06 am
Location: San Diego

Re: MOSFET B+ reducer

Post by lord preset »

I've never tried using a MOSFET but aren't VVRs (Hall, London Power,etc) the practical, proven example of using MOSFETs for reducing B+?

I do know that the MOSFET has to be insulated from the chassis.
User avatar
xtian
Posts: 6990
Joined: Mon Apr 19, 2010 8:15 pm
Location: NorCal
Contact:

Re: MOSFET B+ reducer

Post by xtian »

lord preset wrote:I do know that the MOSFET has to be insulated from the chassis.
matth does not seem to agree with you. He wrote:
how you heatsink the 'fets is different for the CT method vs the VVR method (where you need the mica spacer)
I build and repair tube amps. http://amps.monkeymatic.com
User avatar
lord preset
Posts: 447
Joined: Tue Aug 27, 2013 3:06 am
Location: San Diego

Re: MOSFET B+ reducer

Post by lord preset »

xtian wrote:
lord preset wrote:I do know that the MOSFET has to be insulated from the chassis.
matth does not seem to agree with you. He wrote:
how you heatsink the 'fets is different for the CT method vs the VVR method (where you need the mica spacer)
Well I may not know what I know, if you know what I mean.
User avatar
Structo
Posts: 15446
Joined: Wed Oct 17, 2007 1:01 am
Location: Oregon

Re: MOSFET B+ reducer

Post by Structo »

Isn't the TO-220F package insulated?

Also, does this circuit emit any switching noise?
Tom

Don't let that smoke out!
DonMoose
Posts: 453
Joined: Thu May 29, 2008 10:39 pm

Re: MOSFET B+ reducer

Post by DonMoose »

Just want to mention - that 205W rating is true only if you keep the junction at 25C while you do it.

The line in the datasheet R[theta]ja tells you that, without a heatsink, you'll pick up 62.5C per watt dissipated.
You get to take the junction to 150, so 150-25 = 125.
125/62.5 is TWO watts.

With a heatsink, your thermal resistance R[theta]js is 0.5C/W so 125/5 is 250W, but that assumes a perfect heatsink. The real number will need to add whatever your selected heatsink can do to that 0.5C/W.

Then, a heatsink is only going to help if it's out in the air, not trapped inside an unvented chassis. Whatever is between your device and the heatsink is going to add yet more thermal resistance.

Hope this helps!
R.G.
Posts: 1223
Joined: Tue Dec 02, 2014 9:01 pm

Re: MOSFET B+ reducer

Post by R.G. »

Maybe I can help.
xtian wrote:This idea is often thrown out as an inexpensive way to reduce B+ voltage, but I can't find any practical, proven examples. So I hope to detail my build here for the community.
Good! I've done it, but sometimes I forget to write down exactly what I did, so a third party build is a good way to shake out issues.
1. How much heat will the MOSFET dissipate? Just guessing: if it drops 100 volts at 200mA, that's 20 watts, which matches RG's estimate in the article. I chose a (hopefully) overspec'd N-channel power MOSFET, 900V, 8A, 205W.
Your math is correct. The power in any DC device is the voltage across it times the current through it.
2. How do I choose the Zener? Do I choose a 50v Zener for a 50v drop, and a 100v Zener for a 100v voltage drop?
Nominally yes. Speaking precisely, you have to allow for the turn-on voltage of the MOSFET gate-source. The most common high-voltage power MOSFETs don't start to conduct at all until the voltage from gate to source is more than a few volts. This "cut-in" voltage is like the necessary Vbe forward bias on a bipolar. I tend to use 5V as a cut-in for a generic MOSFET, but they vary. Look up your specific device's datasheet, and you can find the minimum voltage to make the thing conduct. It'll be probably 3-10V. The "5V" estimate is memorialized in the schematic as noting that a 50V zener is equal to a 45V zener plus the voltage in the gate-source.

Looking at the circuit, for voltages lower than the zener voltage, no (significant) current flows in the zener. The voltage from gate to source is therefore nominally zero, and the MOSFET is cut off because its gate source is not large enough to cause it to conduct.

When the voltage rises above the zener voltage current flows in the zener. No current flows in the gate of the MOSFET, which is an almost perfect insulator. So the voltage across the gate-source is equal to the current in the 2K resistor times the resistance, or 2V/ma. When the current reaches the value that turns on the MOSFET, about 2.5ma, the MOSFET starts conducting.

The MOSFET conducts **a lot**. The transconductance (that is, change in drain current per volt of change in gate-source voltage is usually about 1-2 amperes per volt. So when the gate-source hits the cut-in voltage, the MOSFET starts conducting, and conducts just enough to keep its gate-source at the value to keep its current at that level. Confusing? Yep.

Think of it as source follower. The source follows the voltage on the gate, but a few volts lower. So the MOSFET source voltage follows the zener voltage, 45V lower than the drain, plus or minus some small change to change the current. And as voltage on the drain goes up, the source goes up, but 45V plus the Vgs cut-in voltage lower.
3. How do I choose the two resistors shown to match the desired voltage drop?
You don't. Those resistors are constant, not changed for the voltage drop.
4. Can the MOSFET get bolted directly (without mica insulator) to the chassis to serve as heat sink? In other words, can the back of the MOSFET be grounded?
No.

It must be insulated from the chassis electrically, all normal N-channel MOSFETs have the metal tab connected to the drain. If you get a fully insulated MOSFET, sure, bolt it to the chassis.

And now a word from Mother Nature, who wants to say that of Her Laws, thermodynamics are some of the ones She holds most dear. If you get that MOSFET chip inside the package too hot, it will revert to non-useful impure silicon, and stop working. This may cause shower of sparks or the death of other components, or things may simply stop working.

For this to work, you absolutely MUST get the heat out of the chip.

20W in a TO220 is a substantial heat load, and needs a significant heat sink with unimpeded air flow.

If you double this to 40W in a single TO220 device, things get tricky. A stock to-220 needs good, skillful heat sink design and careful implementation to dissipate 40W on a regular basis.

Yes, I know that the spec sheet says it can dissipate - what was it? 205W? Note on the spec sheet that this number assumes a **case temperature** of 25C. You'd have to submerge the device in ice water to keep the case at 25C with 205W coming out of the chip, and even then I would not try it.

The fact is, there is some thermal resistance, expressed as degrees C per watt dissipated, between the chip and that metal flange that's the mounting tab. This is often in the range of 0.75 to 2C/W. If the chip dissipates 100W, and the thermal resistance between chip (junction) and case is 1.2 C/W, then the temperature rise at the chip/junction is 100W times 1.2C/W or 120C. That's above the temperature of the case, and if the case is at 25C, then the chip is sitting at 145C.

Silicon junctions tend to un-diffuse and die at about 150C to 175C. Most power devices made from silicon quote 150 C as the highest junction temperature. Above that, you get smoke.

So you have to add up all of the thermal resistances, from junction to case, from case to sink (this is where the heat sink goo goes) and from sink to ambient air, and add on top of that the ambient air temperature.

The inside of a tube amp often gets to 45C or more. That's 113F. And it can obviously go higher if you're playing a summer concert in Kansas. If 150C is the cliff's edge and you're starting at 45C, you can only stand 105C rise. The thermal resistance from ambient to sink is usually 2-5 C/W. Bolting it to the chassis might get you to 3.5-4C/W if it's thick aluminum. With 40W coming out of the chip, the rise from ambient (45C) to sink is 40*3.5 = 140C. That leaves you only 10C to rise before chip death at 150C.

Ooops. The chip's internal thermal resistance is more than 10/40= 0.25C/W. So this doesn't work.

I blathered all that just to say that the HEAT DISSIPATION on a device burning 40W is a big deal, and can be a deal breaker.

Fully insulated to-220s are often 2 - 3 C/W on their own.

Thermodynamics. Mother Nature *is* a mother.
matt h
Posts: 1224
Joined: Fri May 12, 2006 2:26 am
Location: New England

Re: MOSFET B+ reducer

Post by matt h »

R.G. wrote:
4. Can the MOSFET get bolted directly (without mica insulator) to the chassis to serve as heat sink? In other words, can the back of the MOSFET be grounded?
No.

It must be insulated from the chassis electrically, all normal N-channel MOSFETs have the metal tab connected to the drain. If you get a fully insulated MOSFET, sure, bolt it to the chassis.

While this is true, it's also a reason why, if you do this arrangement another way, you don't need to electrically insulate the metal tab from chassis.

If your drain is the metal tab, and your drain is connected to "real" (i.e., chassis) ground

and your source is connected to your center tap...

ba da bing.

(edit to remove the bad analogy and instead write: if you build your amp with all of your electrical grounds not directly connected to your center tap, you use the mosfet's source as the connection to the CT, thus dropping the CT below earth.)

Not that I can lay a hand on it, this is also how Aiken does it.


edit for clarity
User avatar
martin manning
Posts: 13080
Joined: Sun Jul 06, 2008 12:43 am
Location: 39°06' N 84°30' W

Re: MOSFET B+ reducer

Post by martin manning »

Aiken's circuit ( http://ampgarage.com/forum/viewtopic.ph ... 710#323710 ) and Keen's circuit are essentially the same; both have the drain terminal grounded.
R.G.
Posts: 1223
Joined: Tue Dec 02, 2014 9:01 pm

Re: MOSFET B+ reducer

Post by R.G. »

Yes, you can tie the MOSFET to the chassis electrically if you put the whole thing in the ground side and tie the negative return of the rectifiers to it. That's how the drawing at geofex actually shows it.

I drew that up nearly 15 years ago, and I've had some time to think since then.

There are some issues with bolting it to ground physically, as opposed to just running a wire from the MOSFET drain to the main chassis ground point.

First, it makes the actual heating in the MOSFET even higher if you hook it to the negative side of the rectifiers (which is what the CT is in a FWCT circuit). The rectifiers put large pulse currents into the filter caps, and the equivalent heating of the high pulses is about equal to 1.6 to 1.8 times the DC average current going out of the first rectifier. It makes the heating problem worse. The better way to do that is to have a pre-filter cap, which has a + terminal to the rectifiers and to the actual first grounded filter cap.
The negative side of this first pre-filter cap goes to the rectifier negatives, and the amplified zener goes between that point and the actual circuit ground on the first filter cap. This lets the zener setup handle the average current, not the rectifier pulses and helps keep it cooler.

Second, if you bolt the MOSFET drain to the chassis, you're bolting a signal ground point to the chassis. Good wiring practice dictates that you ground as few places to the chassis as possible, and ideally only one. That forces you to either (1) make the MOSFET the only chassis ground point, (2) move the MOSFET to the place where you want your only chassis ground or (3) insulate it from the chassis. Picking the number and placement of grounding points on a chassis can have implications for hum, noise and stability, so it's nice to have as much freedom as possible for that. Yes, I know that there are long debates about chassis grounding, buss grounding, rail grounding, star grounding, and so on. It's my opinion that insulating the MOSFET from the chassis is worth the trouble in preparation and care, as it buys you some freedom in the list of decisions about where and how many times to ground the chassis.
User avatar
xtian
Posts: 6990
Joined: Mon Apr 19, 2010 8:15 pm
Location: NorCal
Contact:

Re: MOSFET B+ reducer

Post by xtian »

Many thanks, gentlemen! I will be detailing my build here:

http://ampgarage.com/forum/viewtopic.php?p=332352
I build and repair tube amps. http://amps.monkeymatic.com
User avatar
JMFahey
Posts: 252
Joined: Sun Dec 13, 2009 1:39 pm
Location: Buenos Aires - Argentina

Re: MOSFET B+ reducer

Post by JMFahey »

In general, MosFets need insulation.

In just this particular case, the designer chose to ground the MosFet Drain , which happens to be the case, so in this case (and nowhere else) you may dispense with insulation.

Please do not think that can be done in other circuits, specially +V regulators often used to offer adjustable tube amp power.

And grounding drain adds an extra complication, it does not eat 50V out of DC power, but chops 50V out of the peak voltage in a cap charging pulse train, not the same thing.
[img::]https://www.st-andrews.ac.uk/~www_pa/Sc ... 6/fig7.gif[/img]
See that the capacitor is not being charged during the full cycle but only a short time at the peaks.

Roughly, if, say, charge time is 1/4 of the sinewave duration, logic says that current during that short period must be 4X larger than average to compensate.

So your MosFet placed that way, must also stand those huge peaks.

Net effect is that efficiency is lower and MosFet stress is larger than expected.
User avatar
xtian
Posts: 6990
Joined: Mon Apr 19, 2010 8:15 pm
Location: NorCal
Contact:

Re: MOSFET B+ reducer

Post by xtian »

I put a B+ reducer inline with the HT CT in my 18-watt build, which uses reclaimed Hammond iron and a 5U4GB rectifier.

Results, using Aiken's design (attached, on right hand side of PDF) and MOSFET 89K1672.

Code: Select all

Zener   OT CT   drop
none    388v    0
27v     360v    28v
51v     340v    48v
82v     312v    76v
Awesome. Works as designed. No added noise in the amp (and this amp is particularly quiet at idle).

I did not have the MOSFET attached to the chassis or otherwise heatsunk. So with the 76v drop, I limited playing time to 3 minutes, and then measured the MOSFET temp at 250F. I'll make sure it's bolted to the chassis in my build!
You do not have the required permissions to view the files attached to this post.
I build and repair tube amps. http://amps.monkeymatic.com
User avatar
dorrisant
Posts: 2613
Joined: Tue Sep 21, 2010 1:27 pm
Location: Somewhere between a river and a cornfield
Contact:

Re: MOSFET B+ reducer

Post by dorrisant »

Sweet!! Thanks for the update!
"Education is what you're left with after you have forgotten what you have learned" - Enzo
User avatar
xtian
Posts: 6990
Joined: Mon Apr 19, 2010 8:15 pm
Location: NorCal
Contact:

Re: MOSFET B+ reducer

Post by xtian »

More results. I mounted the rig, and added a switch. With a 76v drop and a sine wave running the amp hard, after 10 minutes the temp is only 102F. I turned the volume off, and after another 10 minutes, temp is 103F.
You do not have the required permissions to view the files attached to this post.
I build and repair tube amps. http://amps.monkeymatic.com
Post Reply