Mosfet VVR question

General discussion area for tube amps.

Moderators: pompeiisneaks, Colossal

User avatar
bandade
Posts: 52
Joined: Wed Jul 07, 2010 12:11 am
Location: Dayton, OH

Mosfet VVR question

Post by bandade »

Hi folks, I tried a home built VVR on a cathode biased amp and it worked for a day before it stopped working. It is as if there is no VVR circuit at all and the amp works fine. The zener seems fine and I have tried other zeners with no success

I [u]did not[/u] properly heat sink the Mosfet so I'm pretty sure I burnt it up. I am getting ready to try again with a new mosfet and want to make sure I do it correctly since NTE is not cheap. I am going to heat sink to the chassis and place a mica insulator in between.

Question: After my last mosfet seams to have burnt up, i was able to read the full B+ on the back side of the chip. Is this normal? If it is, I am affraid the insulator with move and I will be sending 350V into my chassis, which would not be good.


Thanks
I have 12V zener diodes though I believe the origional was 6V or 6.5
d95err
Posts: 168
Joined: Sun Jun 18, 2006 6:52 pm
Location: Uppsala, Sweden

Re: Mosfet VVR question

Post by d95err »

It's normal that the MOSFET casing has the full B+. So a mica insulator is necessary. Bolting it to the chassis (with an insulator) should work fine for low or moderate powered amps.

Suitable MOSFETs should not need to be expensive. I used an IRF740 for my Power Scaling circuit. It costs just a few bucks.

You can use pretty much any power MOSFET that has adequate voltage and power handling capacity. Use a hefty margin for the power handling capacity though; these devices were designed for switching, not continous load.
User avatar
ForcedFire
Posts: 34
Joined: Sat Mar 05, 2011 4:35 pm

Re: Mosfet VVR question

Post by ForcedFire »

Is there a particular reason the NTE is used or just everyone copying off someone's schematic? I got some similarly rated MOSFETs for $2 each.
Last edited by ForcedFire on Fri Mar 25, 2011 2:59 am, edited 1 time in total.
NEjoe
Posts: 25
Joined: Tue Sep 07, 2010 1:22 am

Re: Mosfet VVR question

Post by NEjoe »

I find it so much easier to offer suggestions when there is a schematic of what's going on.

ForcedFire I think you nailed the reason, it's just what has been used before, and amp builders go with what works....not necessarily the best choice. Besides once you get into sound then the crazy opinion wars begin filled with the pseudoscientific references about NTE sounding better than IR or ST or whoever.

If you look at what's required, nobody realistically drops more than 300 volts. And current....well B+ for an entire quad of KT88s is still less than 1 amp. The spec that matters for VVR MOSFETs is the power dissipation.

For example, an EL34 has about 140mA max plate current, so a pair would run 280mA (max). And lets say that you want to go from 600V down to 400V.........IxV=P would give us a wattage of 56Watts. However that is with proper heatsink, and 56 watts is a bucketload of heat to get rid of.

Personally I like the offerings from ST since some of them have the protection diodes built in (look for the Z suffix). For example the STP4NK60Z is a 70W, 4 amp, 600V device with diode protection built in. It costs $1.21.
If you feel safer with a bigger power dissipation there is the ST12NK60Z with a 150W rating, 10A capability, 650V rating, $1.89. But, the power ratings mean nothing without proper heatsinks.

Give us a schematic Bandade, we'll fix it.
User avatar
Structo
Posts: 15446
Joined: Wed Oct 17, 2007 1:01 am
Location: Oregon

Re: Mosfet VVR question

Post by Structo »

Isn't NTE just a reseller?

So the component could be any brand really.
Tom

Don't let that smoke out!
teemuk
Posts: 248
Joined: Mon Dec 20, 2010 4:01 pm

Re: Mosfet VVR question

Post by teemuk »

...And who says you have to use just a single MOSFET? There's always an option of fitting two or more in parallel to share the current and dissipated heat.

Plus you can apply all the other tricks in the book: Like fitting a thermal switch to the heatsink to disconnect the B+ feeding the VVR circuit if the temperature ever gets too high. Or monitoring the MOSFET current and limiting it if the current ever rises above safe values.
User avatar
ForcedFire
Posts: 34
Joined: Sat Mar 05, 2011 4:35 pm

Re: Mosfet VVR question

Post by ForcedFire »

NEjoe wrote:.


Personally I like the offerings from ST since some of them have the protection diodes built in (look for the Z suffix). For example the STP4NK60Z is a 70W, 4 amp, 600V device with diode protection built in. It costs $1.21.
If you feel safer with a bigger power dissipation there is the ST12NK60Z with a 150W rating, 10A capability, 650V rating, $1.89. But, the power ratings mean nothing without proper heatsinks.
I went with the STP9NK90Z 160W, couple of bucks at Newark. I haven't installed it yet but I got 2 for half the price of the NTE.
User avatar
jaysg
Posts: 1211
Joined: Tue Feb 07, 2006 11:16 pm
Location: Sandy Eggo

Re: Mosfet VVR question

Post by jaysg »

iirc, Dana specified a part he found to be more reliable than any others he tested.

Is the chassis aluminum or steel? Al is better at spreading heat.
User avatar
UR12
Posts: 1570
Joined: Mon Nov 28, 2005 7:22 pm

Re: Mosfet VVR question

Post by UR12 »

I have went through this quite a few times but I'll try again. I use the NTE part for a couple of reasons. The first being that it is rated at 900v and good up to 275 watts. Since I provide a kit that is being installed in amps anywhere from 5 watts to 50 watts, I need to make sure the mosfet that is included in the kit will work in ANY of those amps. Secondly there is a current limiter built into the VVR and it's operation is dependent on the rating of the zener and the forward resistance of the mosfet's junction (On resistance). You can use any Mosfet you want as long as the On resistance is comparable to the NTE and can get away with using a smaller rated mosfet on smaller wattage amps. You may want to adjust the value of the zener to a 6 volt Zener on amps producing 5 - 10 watts. The NTE part has been extremely reliable with only a hand full of failures on over 1000 installs. Paralleling mosfets when using the VVR may also cause problems with the current limiting. You need to do your homework before you just throw any cheap mosfet in there and expect it to work properly.
User avatar
Super_Reverb
Posts: 188
Joined: Tue Dec 21, 2010 6:28 am
Location: Indianapolis, USA

Re: Mosfet VVR question

Post by Super_Reverb »

Is there a particular reason the NTE is used or just eveyone copying off someone's schematic?
You can use any part that meets the application's requirements, including IGBTs. The design parameters you need to concern yourself with are power dissipation ability, voltage rating (VDss), current capability, and gate voltage range.

If you study the circuit, you'll see you need a Vds or Vce transistor rating a hundred volts or so above your max B+. I'd use a transistor rated at 3-5X expected device current. Max power dissipation will occur at minimum output voltage and max current, because minimum out B+ implies the device will have the largest drain to source or collector to emitter voltage across it. You won't get max current at lowest voltage. Spend a dollar extra and buy a device with higher ratings than you need. You have to heatsink it. It's generally going to be too much power to dissipate in free air.


cheers,

rob
User avatar
ForcedFire
Posts: 34
Joined: Sat Mar 05, 2011 4:35 pm

Re: Mosfet VVR question

Post by ForcedFire »

Good point, the OP doesn't say what amp this is.

Dana, how about a trim pot in series with lug 3 of the VVR pot (between pot and B+ / drain connection) for limiting the maximum B+ available for someone using a transformer that puts out too high a voltage? Useful for someone using a transformer spec'ed for tube rectifier and using solid state rectifier ending up with too much B+.
User avatar
UR12
Posts: 1570
Joined: Mon Nov 28, 2005 7:22 pm

Re: Mosfet VVR question

Post by UR12 »

ForcedFire wrote:Good point, the OP doesn't say what amp this is.

Dana, how about a trim pot in series with lug 3 of the VVR pot (between pot and B+ / drain connection) for limiting the maximum B+ available for someone using a transformer that puts out too high a voltage? Useful for someone using a transformer spec'ed for tube rectifier and using solid state rectifier ending up with too much B+.
Yes, you could do that if you like. My new VVR3 boards for
fixed bias amps now has a place to add a resistor to the top of the
pot. I ship them with a jumper in that position on the board. You can
just clip the jumper out and add a resistor of the proper value
to lower the B+ to what ever you want.
User avatar
ForcedFire
Posts: 34
Joined: Sat Mar 05, 2011 4:35 pm

Re: Mosfet VVR question

Post by ForcedFire »

Ah cool, good idea. I still haven't tested the transformer I'm using on my current build so I don't know if I'll need to limit my max B+. I think I will install a trimpot just in case.
User avatar
UR12
Posts: 1570
Joined: Mon Nov 28, 2005 7:22 pm

Re: Mosfet VVR question

Post by UR12 »

Super_Reverb wrote:
Is there a particular reason the NTE is used or just eveyone copying off someone's schematic?
You can use any part that meets the application's requirements, including IGBTs. The design parameters you need to concern yourself with are power dissipation ability, voltage rating (VDss), current capability, and gate voltage range.

If you study the circuit, you'll see you need a Vds or Vce transistor rating a hundred volts or so above your max B+. I'd use a transistor rated at 3-5X expected device current. Max power dissipation will occur at minimum output voltage and max current, because minimum out B+ implies the device will have the largest drain to source or collector to emitter voltage across it. You won't get max current at lowest voltage. Spend a dollar extra and buy a device with higher ratings than you need. You have to heatsink it. It's generally going to be too much power to dissipate in free air.


cheers,

rob
Here is a cut and paste from this forum from a few years back when this was all discussed at length by PRR
PRR wrote:> deal with the heat problem

Total heat in chassis falls. (Unless you done something wrong.)

Heat in power tubes falls. But we have a new heat source, the MOSFET/resistor voltage-dropper. In a tight "authentic" chassis, finding a place for this may indeed be a problem.

> no heat problem until you get above 50 watts.

You did end up with a finny-thing hanging off the end. You do point out that it is overkill for an 18W.

Take an "18W". Cathode-biased, it'll idle around 36 Watts heat. It may run 300V or 350V, but take 300V and 120mA as an example.

18W out - 300V 120ma = 36W tube dissipation - 0V 120ma = 0W regulator diss - 36W total
9W out - 210V 84ma = 18W tube dissipation - 90V 84ma = 7.5W regulator diss - 26W total
4W out - 150V 60ma = 9W tube dissipation - 150V 60ma = 9W regulator diss - 18W total
2W out - 105V 42ma = 4.4W tube dissipation - 195V 42ma = 8W regulator diss - 13W total
1W out - 75V 30ma = 2.3W tube dissipation - 225V 30ma = 6.8W regulator diss - 9W total

The maximum heat in the regulator is one quarter the power in the output stage when working full-up. For a cathode-biased amp, this is pretty-near the idle heat; round-up about 20% because we let self-bias amps slide a bit into Class AB. For a fix-bias amp, look-up the operating conditions for a similar application. 2*6L6GC ala 5F6A will idle near 400V 100mA but roar to 400V 220mA, so at half-voltage there could be up to 200 110mA or 22 Watts heat in the regulator.

The maximum heat happens at half-voltage, which will give something less than quarter power output.

The heat in the regulator is fairly constant for all outputs from 50% power to 5% power.

9W heat is not a big problem for modern MOSFETs on a good chassis or aluminum flange. At 22W regulator heat you do have to think where to put the large surface area (or fan) needed.

But the total heat in the box falls. Adding power reduction will not blister your Tolex any more than before, and generally less if you actually use the power reduction. Unless you are "forced" to put the heatsink too close to the cabinet.

> a switching supply for tube amps. ...have a Variable Voltage power supply where you are varying the AC and not the DC.

It's an interesting thought. The standard PC power supply has the Watts we need. And it is "variable", though wired to vary to keep itself at a multiple of the 2.5V reference. Fiddle a resistor you can swing it several volts. To get 400V instead of 5V you would have to re-wind the HF transformer. Copying the input winding as a new secondary winding would put you in a ballpark. (That's where I get stuck: there is no way my older eyes and fat fingers can re-wind those dinky cores.) You'd need fast high-voltage rectifiers, but they can be had. You have to figure out the regulation resistor ratios for your new voltage. And for variable power, it may get tricky. Some of these switchers won't turn-down very far without motorboating. Also it seems sweet to take your heater power from a jazzed 5V output since this is so much lighter than a 50/60Hz heater transformer, but one regulator runs all the outputs so we can't hold 6.3V on one while varying 400V-40V on the other. (However surplus 12VDC switchers are dirt-cheap.)
NEjoe
Posts: 25
Joined: Tue Sep 07, 2010 1:22 am

Re: Mosfet VVR question

Post by NEjoe »

The NTE2973 is a perfectly fine part, but it's overkill in every single direction for many, if not most, situations. I'll concede that if you're selling a kit you have to try and prepare for anything the general public might do (even when you say not to) and that will likely lead to using the most indestructible part possible. However, that's an entirely different criteria from what I'm doing. And it sounds like the original poster might need a price break from the NTE part.

Ridiculous overkill is fine, and if the NTE cost a few pennies more I might give it a shot. However, when that overkill comes with a fairly hefty pricetag.... $10 versus $2 or $3 for any appropriately sized device....well I'm not interested. Besides, the NTE2973 is overpriced. Parts with even more power dissipation capability (350W for a STW15NK90Z) and equal or better specs still cost less than $7.
I also really like the built in back to back protection diodes the ST offers, no worries about picking 12V or 6V diodes and no polarity concerns since both directions are covered.

But that's just my opinion, worth exactly what you paid for it. 8)
Post Reply