Zener diode temperature question

[osing]

Howdy,


I have a Tweed Bassman style amp I built where I have lowered the B+ voltage using five 10V 5W zener diodes between the power transformer center tap and ground. With these, I get a plate voltage of ~430V or so using 5881 tubes. I noticed that these diodes were getting very hot, so I added two heat sinks, one across the bottom and one across the top of the diodes, attached with adhesive thermal compound (I used something called Arctic Alumina Thermal Adhesive 5g). Anyway, the heat sinks are getting so hot that I burned my finger when touching. After playing the amp with the volume maxed out using the 5881 tubes, I measured the temperature of the heatsink to be 250F or so (using a thermocouple and my DMM). Using 6L6GC tubes, I measured 300F or so after having played at full volume. At idle, before playing the amp at all, I measured 170F which is also quite hot to the touch!


Concerned that I might be overtaxing the diodes, I soldered a 1 Ohm resistor in series with the CT and measured the voltage across it and got (I = V/1 Ohm) ~75mA of current at idle with the 5881 tubes and close to 100mA with the 6L6GC tubes. While playing aggressively and at full volume, with my meter set to DC, I saw voltages around 200mA-250mA or so with the 5881 tubes and a little over 300mA with the 6L6GC tubes. I don't know how accurate these readings are, but I got similar readings whether the meter was set to AC or DC. These are 10V zeners and probably drop a little less than that, but at 10V the dissipation at idle would be ~0.75W for the 5881 tubes and ~1W for the 6L6GC tubes. At max volume and aggressive playing, the dissipation would be ~2.5W with the 5881 tubes and ~3W with the 6L6GC tubes. I assume this should be within a comfortable margin of the 5W diode rating. I looked up the specs for the zener diodes I bought, and they do say max operating temp of 200C which is close to 400F.


My question is, has anyone here used zener diodes in this manner to lower B+ and if so, have you ever noticed how hot they get? Do the currents above seem about normal for this type of amp?


Glenn

[MoparWade]

Why not use resistors?

[osing]

My understanding from what I have read is that resistors will induce sag whereas zener diodes will not. I have not tried resistors, so I cannot speak to how much of a difference that sag makes (good or bad, I suppose, depending on what you are looking for)

[MoparWade]

My thinking would be that we use resistors and caps to create nodes of decreasing voltage for the various valves in an amp so I would apply that same concept to what you're doing. I am by no means an expert though so hopefully someone who is will chime in.

[MoparWade]

Also why on the CT? It seems like you could calculate the voltage drop you wanted from the current which you know and the 430 volts you want at the plate. Assuming you want a 50V drop then you're at 480V. So 50=(0.075)R gives 667 Ohm's. A 670 ohm 4W resistor on your B+ should work.

[HotBluePlates]

... I measured the temperature of the heatsink to be 250F or ... 300F or so after having played at full volume. ... At max volume and aggressive playing, the dissipation would be ~2.5W with the 5881 tubes and ~3W with the 6L6GC tubes. I assume this should be within a comfortable margin of the 5W diode rating. I looked up the specs for the zener diodes I bought, and they do say max operating temp of 200C which is close to 400F. ...

I haven't used zeners for voltage adjustment, so I can't tell you how hot is "normal". But just like with resistors, the closer the actual dissipated power in the zener gets to the rating of the device, the hotter it gets. If you were using 10w zeners, they would run cooler. If you used even higher-wattage zeners, they would get cooler still...

The currents you were measuring sound about right for peak current draw at max output for a ~50w amp.

Why not use resistors?

Ohm's Law is why.

Volts = Current * Resistance. If we used a resistor to drop voltage, then there is little voltage drop when current is small (idle) but very big voltage drop when current is big (full roar). So the voltage drop is non-constant.

A zener used within its ratings gives a constant voltage drop. So the lowered supply voltage is consistent at idle and at full roar.

... Assuming you want a 50V drop then you're at 480V. So 50=(0.075)R gives 667 Ohm's. A 670 ohm 4W resistor on your B+ should work.

75mA * 50v = 3.75w. A 4 watt resistor will get screaming hot, though it shouldn't burn up. The common rule of thumb is use a resistor double-wattage the expected dissipation, so 7.5w and probably rounded up to a standard 10w part.

That's fine for idle, but osing pointed out the current is as much as 300mA at full-tilt. 300mA² * 670Ω = 60.3 watts. Your 4w resistor (or even my 10w resistor for the idle current) just went "poof"... And before it did, it tried to drop 300mA * 670Ω = 201v instead of the original 50v.

Also why on the CT? ...

The idea is consistent reduction of voltage, right at the power transformer where the original a.c. voltage is provided to generate the B+ voltage. And the reduction is created by shifting the PT winding's relationship to the reference point of "ground".

The PT's voltage is from the end of the winding to ground at the center-tap for a center-tapped high voltage winding. The zeners move the center-tap 50v "below ground" so the CT appears to be at -50v relative to ground. But the voltage of the winding from CT to the end of the winding hasn't changed. When you measure voltage from end of the winding to ground, it appears to be the winding voltage minus 50v.

Hypothetical example:
You have a PT with a high voltage winding of 340-0-340v. With a solid state rectifier, you'd get 340v * 1.414 = ~481vdc (minus diode drop). The winding looks like
340v (winding output to rectifier, filter cap)
0v (grounded CT)

You add a zener string which moves the CT 50v "below ground". The winding now looks like
290v (winding output to rectifier, filter cap)
0v (grounded filter cap at rectifier output)
50v (voltage drop across zener string, with CT on non-grounded side)

Total winding voltage is still the same 340v it was before. But the filter cap sees a rectified 290vac, for a rectified 290v * 1.414 = 410vdc (minus diode drop).

And with the zeners, the rest of the amp sag/interaction at high volume is unchanged over what you'd have if you just wired it normally but used an actual 290-0-290v PT.

[MoparWade]

See I knew an expert would come along if I made a big enough hash of things. It all makes sense now.

[2deaf]

See I knew an expert would come along if I made a big enough hash of things.

Yeah, sometimes you gotta flush them out.

[2deaf]

So now that you know the zeners will get hot, have you considered any options?

If you have more 10V 5W zeners, you could put a string of five on each end of the winding and ground the center tap.  Still gets hot, but not as hot since each zener is only conducting half of the time.

10W zeners seem to be a little scarce and a lot pricey.  You could buy a bunch of lower voltage 5W zeners and string those together.

It seems like a MOSFET VVR with a fixed voltage divider on its gate would be a better solution for this much wattage.

[NEjoe]

A few thoughts......essentially you're trying to make 50 Volts disappear.....something has to get hot.

No matter how you do it, the energy of 50 Volts times 300mA has to go somewhere, turned into heat by some components.

15 Watts worst case, spread over 5 diodes. 

Sounds like it's working to me, I'd say you're fine.  You're under the max Watts and temp for the components.   

There's other ways to do it, but so what?  Sure you can rig up a slicker MOSFET follower for $4 in parts (and make it variable), but it will still kick out the same amount of heat.  I will say the mosfet would get you a better heatsink platform.

BTW for effective and cheap thermal grease, use antiseize compound, the silver colored slop is good for 1600 degrees or so.  Just a little, since you're only trying to fill all the gaps to get as much contact as possible with your heat sink.

[HotBluePlates]

A few thoughts......essentially you're trying to make 50 Volts disappear.....something has to get hot.

No matter how you do it, the energy of 50 Volts times 300mA has to go somewhere, turned into heat by some components. ...

For the most part, yes.

He could have used a bucking transformer to lose the 50v with very little waste heat. But that's because the voltage & current would be out of phase.

[2deaf]

Sure you can rig up a slicker MOSFET follower for $4 in parts (and make it variable), but it will still kick out the same amount of heat.

I don't think one of those MOSFET big boys is going to get very hot dropping 50V while bolted to the chassis.  If it were variable, it would get pretty toasty in the middle third of its range, but run cooler at both extremes.

[osing]

I figured I would come back to my thread and see now that several folks chimed in - thanks for all the replies! I guess I forgot to select "notify me of replies"! Anyway, I got busy with other things and finally got back to my amp project again yesterday. Although the 5W rating on the zener diodes at first seems sufficient, there is a note in the spec sheet that suggests you have to de-rate the 5W rating by 40mW/C above 25C. If I understand this correctly, if the temperature of the device reaches 300F, which is ~150C, then the 5W rating should be reduced to 0W (150C is 125C above 25C, 125 *0.040W = 5W). Then again, it says max operating temp of 150C, so I am not sure what to make of this other than that I don't feel good about having measured the temperature of the heat sink I had attached to the zeners to go as high as 300F!


I came up with a different heat sink solution. Essentially, I took a piece of aluminium, drilled holes for the 5 zener diodes, used thermal adhesive to make sure they have good contact with the aluminum, then applied heat sink thermal pads under the aluminum bar and screwed it to the chassis. I then added standoffs to the screws. The leads of the zeners were bent up in the air and then finally a small turret board was placed onto the leads and standoffs. The leads were then soldered to the turrets. After trying it out, it seems to dissipate the heat much better than my previous heat sink approach (it should given that the chassis now serves as a heat sink) and I did not measure temperatures much above 100F so far, which is great!


If repairs are ever needed, the zeners would need to be drilled out, so that is not ideal, but I am hoping this solution will work and no repairs are needed anytime soon :-)


Attached are three images - one of the aluminum bar with the glued in zeners, one of the board from above (the zeners and aluminum bars are underneath it, please ignore the melted insulation of my heater wires...), and the zener specs.


Anyway, I figured I would just let y'all know how this all worked out.

[tubeswell]

... measured the voltage across it and got (I = V/1 Ohm) ~75mA of current at idle with the 5881 tubes and close to 100mA with the 6L6GC tubes. While playing aggressively and at full volume, with my meter set to DC, I saw voltages around 200mA-250mA or so with the 5881 tubes and a little over 300mA with the 6L6GC tubes.

Each 10V zener is capable of dissipating 5W. Assuming the 300mA draw from the 6L6GCs running flat out, each zener sees 10V x 0.3A = 3W. This would be a little on the high side (2.5W would be optimal average maximum current draw). However, the amp doesn't see 300mA on the HT most of the time, so they are probably ok if there is adequate ventilation/heatsinking.


 Switching to lower voltage 5W zeners would bring the dissipation down a bit.

[PRR]

Since the Zeners are in the spike-wave part of the rectifier, the RMS (heat) current will be FAR higher than the DC current.

Sim says for 200mA DC, the RMS current of the spike-wave in the diodes is near 400mA.

10V*400mA is 4 Watts.

A "5W" part will not carry 4 Watts without a tight clamp and water-cooling.

Modern Silicon is pretty tough. But I think you will have "wear out" long before the 100,000 Hours it should be good for. Maybe not even 1,000 hours.

If you can find a 150W P-type MOSFET, you can bolt the Drain tab to a spreader plate on the chassis _AT_ your main filter cap ground connection, without insulation (do need grease!). A small 45V Zener and a 1K Gate-Source resistor makes it a 50V 50+W Zener.

[osing]

Thanks guys. I guess I will spot check the voltages and bias every once in a while and see if anything changes. I don't know if the Zeners wear out gradually or simply fail, but if I don't hear a pop or see smoke, and if the amp still works after they wear out or fail, the voltages and output tube bias will be off. I only get around to playing an hour or two a week, so I suppose premature failure in terms of component life might still take a while to manifest itself. However, if that does happen, it sounds like there are some other options to consider - thanks for the suggestions!

[tubeswell]

Thanks guys. I guess I will spot check the voltages and bias every once in a while and see if anything changes. I don't know if the Zeners wear out gradually or simply fail, but if I don't hear a pop or see smoke, and if the amp still works after they wear out or fail, the voltages and output tube bias will be off. I only get around to playing an hour or two a week, so I suppose premature failure in terms of component life might still take a while to manifest itself. However, if that does happen, it sounds like there are some other options to consider - thanks for the suggestions!

If they go open when installed in series with the CT and ground return (in the way that they are in your amp), all that will happen is the amp will go into standby mode (i.e. - same way that some early Fender amp's standby switches were wired in)