Help with dropping B+ with Zeners

[Dr Tony Balls]

Hi all...on a build where my PT is a bit over spec'd for the job. Its a sort of 4xEL84, cathode biased, plexi-type amp and the PT i'm trying to use is spec'd for 290-0-290 on the secondary. With full wave rectification, and the bias at 92% dissipation i'm seeing 367 VDC on the plates. That's a bit high so I thought i'd try the Zener diodes on the PT center tap trick. I'm using four 5W/12V Zeners (1N5349B) in series on a tag strip to try and drop 40 something volts or so. Its working to drop the voltage just fine. Initially I was seeing 322 VDC on the plates, and with a re-bias i'm seeing 310 VDC on the plates at 88% dissipation. The problem i'm getting is the diodes are smoking!

The current draw on the HT ought to be around 220mA. If each diode is dropping 12V then (12 x .220) then each diode would be hitting 2.64W. I'd think at 5W spec i'd be fine for this, but am I incorrect?

Here's pics of my tag strip:






And i'm using this approach:




Any ideas? I'm not heat-sinking anything at current because I figured i'd be safe with double the power rating I needed but maybe that's incorrect.

[johnnyreece]

Got a picture of your rectifier circuit?

[sluckey]

Are they literally smoking? Or are you just saying that when you really mean they are too hot to touch? 2.5W dissipated in that small area will likely put a blister on your finger.

[Dr Tony Balls]

Got a picture of your rectifier circuit?

Yeah its very basic just two diodes:




I'll also add this.... i broke out the infrared thermometer to try and get some reading and while it's hard to get good results on small diodes, I was seeing different results on different diodes. Going CT > D1 > D2 > D3 > D4 > Gnd I was seeing D1 and D3 at ~180-200F but D2 and D4 were at room temperature. D4 I could see being *somewhat* lower because of its connection to the chassis but thats a lot, and on D2 it doesnt make sense.

Are they literally smoking? Or are you just saying that when you really mean they are too hot to touch? 2.5W dissipated in that small area will likely put a blister on your finger.

Yes literally smoking.

[Stevem]

I would take full advantage of that excess voltage and a class A output stage like you have ( no sag does not matter ) and add another stage of filtering and a dropping resistor to get you down to 300 volts.

[johnnyreece]

Got a picture of your rectifier circuit?

Yeah its very basic just two diodes:

Well, there goes my idea...I thought maybe you were using a bridge rectifier while using the CT. This is obviously not the case.

[SoulFetish]

I would be surprised if RG doesn't jump in on this topic and express that you can use his zener/mosfet combination (from mosfet follies ver...1?) to great effect for your purposes. It is simple, effective, inexpensive, and I used it in a amp I built incorporating 2 EL84s and 3 12AX7s.
For that build, I had a PT on hand that was designed for 4 EL84s, handful of 12AX7s, and a winding for solid state switching. The loaded secondary was elevated about 35V or so higher than I wanted, so I used his "mosfet B+ reducer" described below to bring down my B+ --
http://www.geofex.com/Article_Folders/m ... tfolly.htm
(scroll down)
He shows it used in a full wave (2-phase) rectifier, like the one you are using, but it can also be equally used in a bridge rectifier circuit.
Dude, FETs & zeners, are super useful tools. If you know how they work, they are just a clever idea away from saving the day.

[Dr Tony Balls]

I would be surprised if RG doesn't jump in on this topic and express that you can use his zener/mosfet combination (from mosfet follies ver...1?) to great effect for your purposes. It is simple, effective, inexpensive, and I used it in a amp I built incorporating 2 EL84s and 3 12AX7s.
For that build, I had a PT on hand that was designed for 4 EL84s, handful of 12AX7s, and a winding for solid state switching. The loaded secondary was elevated about 35V or so higher than I wanted, so I used his "mosfet B+ reducer" described below to bring down my B+ --
http://www.geofex.com/Article_Folders/m ... tfolly.htm
(scroll down)
He shows it used in a full wave (2-phase) rectifier, like the one you are using, but it can also be equally used in a bridge rectifier circuit.
Dude, FETs & zeners, are super useful tools. If you know how they work, they are just a clever idea away from saving the day.

Right, I know there are other options which I may get to eventually but the thing puzzling me is why is the option i'm trying not working when it seems to work quite suitably for others. The math checks out, I think.

The only thing I can think is that maybe these are cheap/fake diodes? They've been in a parts bin that I inherited for years, and I dont know where they came from. They're labeled 1N5349B and have an On semiconductor logo but perhaps they are phony? That might explain why some seemed to be overheating and others didnt.

[martin manning]

Are the ones at room temp shorted?

[R.G.]

OK - I'll jump in.

Martin's on it. Are one or more shorted?

Edit: Just after I hit submit more occurred to me.
Have you measured the voltage across each diode? That will tell the tale for whether some are shorted or not, or maybe fake/mislabeled. Also, have you actually measured the current through the diodes?

[Dr Tony Balls]

Are the ones at room temp shorted?

I had thought of that, but figured a shorted diode wouldnt drop any voltage. And since i was seeing ~48V drop from four 12V diodes that would indicate all are working. I tested them just now, though, and all read at .638V forward bias voltage drop. All the others I have of the same diode read the same, as well.

OK - I'll jump in.

Martin's on it. Are one or more shorted?

Edit: Just after I hit submit more occurred to me.
Have you measured the voltage across each diode? That will tell the tale for whether some are shorted or not, or maybe fake/mislabeled. Also, have you actually measured the current through the diodes?

I turned the amp back on and measured the voltage across each diode and they all initially measured an 8V drop at startup with the full string measuring 32V. The string reads 0 - D4 - 8 - D3 - 16 - D2 - 24 - D1 - 32

After a minute or so D1 and D3 heat up to ~200F and the readings change and i'm not getting an even drop across each diode. The string reads 0 - D4 - 9 - D3 - 14.5 - D2 - 20.3 - D1 - 25

I have not measured the current through each diode yet. I'll see if I can give that a go after lunch.

[R.G.]

Hmmm.
200F is about 100C, which is hot but OK for a power semiconductor. Odd that they are changing voltage. Must be something with the current.

You only have to stick a current meter in series on either end of the string to see the current in all of them.

This is way too early to suggest before you find out the current, but one thing that can keep the dissipation down in a zener string is to parallel it with a resistor so the resistor drops the right amount of voltage at the minimum current flow, and the zeners take over for any current between the minimum current and the maximum. As I say, it's premature to think about right now, but my mind just keeps jumping to circuits.

[sluckey]

I tested them just now, though, and all read at .638V forward bias voltage drop.

That's correct for forward bias. But we are interested in the reverse bias voltage, which is the rated voltage for the diode, in this case, 10V. Your meter probably can't test that without applying an external voltage source.

After a minute or so D1 and D3 heat up to ~200F and the readings change and i'm not getting an even drop across each diode. The string reads 0 - D4 - 9 - D3 - 14.5 - D2 - 20.3 - D1 - 25

Probably gonna need some 10W zeners with a heatsink.

I have not measured the current through each diode yet. I'll see if I can give that a go after lunch.

The current is the same through every diode since they are connected in series. Just disconnect the center tap lead and insert your amp meter between the diode stack and the center tap.

[Stevem]

I have a question.
Are we not introducing some level of zener switching noise into the picture here?

If so then if our goal is low noise, be it 60 hz, 120 hz or switching hash noise, then why go this route with the zener’s just because we can?

[R.G.]

Are we not introducing some level of zener switching noise into the picture here?
If so then if our goal is low noise, be it 60 hz, 120 hz or switching hash noise, then why go this route with the zener’s just because we can?

I don't think so, although I haven't dug into it from that direction.

Just mulling it over in my head, the zeners in this setup are always "off" from their internal perspective. They just go into avalanche mode where the voltage across them pulls charge carriers across the reverse diode junction. This makes for nearly white "hiss" but not so much low frequencies. The zeners don't slam off, they just drop below avalanche voltage.

Diode switching noise is caused by the reversal of voltage across a conducting diode sweeping out the charge carriers in the conduction area and suddenly slamming off in nanoseconds, making an abrupt current transient that makes the parasitic inductance and capacitance of the wiring ring, often at RF frequencies. This makes a squark of RF that the circuit rectifies into a blip of audio at twice line frequency.

Avalanche hiss is more easily suppressed by the filter caps, being higher frequency. The RF blips are harder for the main capacitors to suppress because the electrolytic filter caps have bigger internal ESL; their impedance actually goes up with higher frequencies. This is one place where paralleling electros with smaller ceramic caps might help, although using fast, soft-off rectifiers like FREDs would just not generate the RF to start with.

I'd have to dig deeper, but my intuition says that the zeners won't add noticeable noise. But your intuition is good - be suspicious of add-on fixes. They could cause other issues and side effects.