Extremely hot resistor

[warpie]

ok, thanks jdbakker, I think it starts to make sense now.
So, in this case the power is:

294V - 272V = 22V

(22V x 22V) / 470 = 1.02W

is that correct?

thanks again for your help 

w.

 

[syn]

Yes, that is correct. But, in my case, there are some weird, extra things to consider as I'm using 15W part,
and it is still too hot to touch.

 

[warpie]

phew!  I spent hours trying to figure it out..  ;D

thanks guys
w.

 

[jdbakker]

in my case, there are some weird, extra things to consider

Nothing weird here. Walrus asked earlier:

What type of meter are you using? The only thing I can think of is the ripple is enormous and actual peak to peak volts is much more than 22 volts hence higher dissipation. Have you got a scope to check the waveform on C14?

If you used a regular (digital or analog) voltmeter, the reading you got on the left side of R33 is not just meaningless, it's misleading. Almost all multimeters show an average value, whereas for power calculations you need the RMS of the voltage across the resistor.

I'm using 15W part, and it is still too hot to touch.

Why is that bad or exceptional?

JDB.

 

[syn]

in my case, there are some weird, extra things to consider

Nothing weird here. Walrus asked earlier:

What type of meter are you using? The only thing I can think of is the ripple is enormous and actual peak to peak volts is much more than 22 volts hence higher dissipation. Have you got a scope to check the waveform on C14?

If you used a regular (digital or analog) voltmeter, the reading you got on the left side of R33 is not just meaningless, it's misleading. Almost all multimeters show an average value, whereas for power calculations you need the RMS of the voltage across the resistor.

I'm using 15W part, and it is still too hot to touch.

Why is that bad or exceptional?

JDB.

Weird part is that after checking many times, after replacing the components (good quality or at least decent, like panasonic, nichicon, TI etc) the situation stays the same. If you look at Jacob's schemo, you will notice that there is 2W rating for this resistor. I'm using 15W part, which after a couple of minutes becomes too hot to touch, for more than a couple of sec. There is ~46mA through 470ohm=~1W. And there is this huge ripple... I find that weird...
Somehow, I did not notice Walrus 's question. Sorry.  I'm using "true rms" (UNI-T/ UT60H) meter, that I used on every single build up to now, with predictable results.
It is not bad or exceptional on it's own, but it is bad and exceptional to have high ripple and too-hot-to-touch parts that can not be controlled or their behavior  understood (at least by me).

 

[keefaz]

1 watt dissipation or so is hot to touch, and even a 15 watt resistor will produce heat maybe with a longer time than a 2 watt part tough
For the ripple, did you measure it with your audio circuit connected to the G9 PSU ?

For my part I used the G9 PSU in a stereo phono tube preamp with success, I mean it is very quiet

 

[jdbakker]

Weird part is that after checking many times, after replacing the components (good quality or at least decent, like panasonic, nichicon, TI etc) the situation stays the same.

There are two issues here:

(1) Your meter does show average levels and not RMS for DC measurements, as far as I can tell from the manual.
(2) You seem to believe that it's a problem for a power resistor to get hot to the touch.

Ad 1: I had a brief look at the UT60H manual, and the RMS measurements only applies to the AC setting, not the DC range. Without further specification I have to assume that your meter is showing an average value. While the capacitor voltage has little enough ripple to be considered DC (I'd guess ~3Vpp ripple for a 150u cap and 46mA), the voltage on the bridge rectifier is only high enough to have a current through R33 for part of the cycle. Back of the envelope calculations show that the bridge is conducting less than half of the AC cycle, and RMS current through the resistor is probably somewhere between 65 and 80mA, for a dissipation between 2 and 3W.

(I'm a bit worried that your meter's manual has such strong warnings about measuring voltages more than 60VDC/30VRMS above ground. Not sure if that's to please the lawyers or an indication of a bigger issue.)

If you look at Jacob's schemo, you will notice that there is 2W rating for this resistor.

Ah, but the voltage at his bridge rectifier is a few dozen volts lower than yours, and current consumption is lower too. In your case I would have preferred a 200-220V secondary, BTW.

I'm using 15W part, which after a couple of minutes becomes too hot to touch, for more than a couple of sec.

And why is that a problem?

There are two ways to make a resistor able to dissipate a lot of power: lower its thermal resistance to the ambient air, or increase its max working temperature. If you can keep your finger on it for a few seconds without getting a blister or losing skin, I'd say it's running at ~70 degrees, or 45-50 above ambient. That's not abnormal; it's not uncommon for power resistors to run at or over 200 degrees at full rated dissipation.

And there is this huge ripple... I find that weird...

Where do you see this huge ripple? Like I said I would expect ~3Vpp ripple at the input of the regulator.

JDB.

 

[syn]

JDB

      thank you very much. Everything clear. The ripple I see is when I plug this compressor to the RMAA,
I have ~-50dB peak at 100Hz (unweighted ref. ~1.22VAC). The circuit otherwise measures decent.
"Star" grounding, all grounds separated (HTgnd,6.3Vgnd,Audio gnd) meet only at star point and then go to "mains" gnd.
Best

 

[syn]

(2) You seem to believe that it's a problem for a power resistor to get hot to the touch.

I don't think it is a problem for it to get hot to the touch, I think that ripple is having something to do with that
resistor getting too hot, or the other way around.

 

[Kingston]

That 100hz peak might not have anything to do with B+.

What's your heater PSU like?

what happens to the 100hz peak when you switch off heaters while B+ is still on? You can do a quick check like this without destroying any tubes (but not longer than 5-10 seconds).

you mention HTgnd,6.3Vgnd,Audio gnd

what do you mean with the separate audio ground? Besides some shields and XLR pin1's to case/star ground, your audio ground should be the same as HTgnd (sometimes called B-). "audio ground" flows to the PSU. There should be no separate connections to star ground anywhere inbetween.

 

[jdbakker]

What Kingston said.

I think that ripple is having something to do with that resistor getting too hot, or the other way around.

Not likely. R33 affects higher harmonics of the mains by forming a low-pass filter with the input cap, and it dissipates power that would otherwise heat the IC regulator, but the level of the 100Hz fundamental is dominated by load current and input capacitor.

JDB.

 

[Kingston]

Also, let us know the RMS VAC measurement at your PSU output (the point where you measure 239VDC). I'm very familiar with this particular type of PSU, and a reasonable quality multimeter should show a value less than 1mVAC. My fluke for example says 0.4mVAC when there's no ripple (but can lie in certain cases when ground is polluted).

 

[syn]

Heater PSU= LM 350 T + surrounding stuff, set to 6.3VDC.
For Audio gnd i ment "some shields and XLR pin1's", sorry i was not clear enough.
I'll see to the "no 6.3VDC test".
Good, now I know, that  hot resistor and ripple don't have anything to do with each other.

Thank you very much

 

[syn]

Also, let us know the RMS VAC measurement at your PSU output (the point where you measure 239VDC). I'm very familiar with this particular type of PSU, and a reasonable quality multimeter should show a value less than 1mVAC. My fluke for example says 0.4mVAC when there's no ripple (but can lie in certain cases when ground is polluted).

O.K I will.

 

[Kingston]

Good, now I know, that  hot resistor and ripple don't have anything to do with each other.

Some notes about R33.

You could even remove it completely. You'll just be heating up the regulator a bit more this way, but probably affecting performance very little.

Or if you want to stick with R33, at least add another cap before it (from rectifier bridge output to ground). That way you'll present only like 0.4VAC to the regulator, and with the already added R37+cap at output you should have a *very* stiff PSU in your hands.

Heater PSU= LM 350 T + surrounding stuff, set to 6.3VDC.

Let us know the VAC RMS at heater output as well.

 

[syn]

It stops @ 0.01VAC
than slowly fails to 0.009
than slowly fails to 0.008
that is over around 20sec.

 

[jdbakker]

Or if you want to stick with R33, at least add another cap before it (from rectifier bridge output to ground). That way you'll present only like 0.4VAC to the regulator, and with the added R37+cap at output you should have a *very* stiff PSU in your hands.

...at the expense of higher radiated harmonics.

If you want to experiment with this, I'd suggest using a resistor in series with this cap. A good starting point would be 1/10...1/4 of R33, or 47..120R.

JDB.
[this would be a good place not to use an expensive low-ESR capacitor]

 

[Kingston]

...at the expense of higher radiated harmonics.

Maybe in a passive PSU I would take it into account.

But our regulator should handle all these gracefully, shouldn't it? Were not talking some exotic 2mhz crud, but low harmonics all perfectly in the range of the chips ripple rejection. And what the regulator doesn't handle the last "safety RC" will.

 

[jdbakker]

...at the expense of higher radiated harmonics.

Maybe in a passive PSU I would take it into account.

But our regulator should handle all these gracefully, shouldn't it? Were not talking some exotic 2mhz crud, but low harmonics all perfectly in the range of the chips ripple rejection. And what the regulator doesn't handle the last "safety RC" will.

The regulator eats most of the conducted harmonics, yes. Radiated harmonics (transmitted by the transformer and the loop formed by transformer-bridge-cap) end up in insufficiently shielded loops and audio transformers (for the magnetic component) and on nearby conductors (for the electric field). The stronger HF components in the charging current of a cap connected to the bridge can also pollute ground if you're not careful, especially in a layout that didn't take this cap into account to begin with.

(You might say that a large fraction of linear supplies have a cap directly connected to the bridge rectifier, and thus suffer the same problem. Can't argue with that, but if you're adding a $1 cap to reduce ripple you might as well add a $0.01 resistor to reduce the cap's side effects.)

JDB.

 

[syn]

Heater VAC RMS have very similar behavior.
0.011VAC
than slowly fails to 0.010
than slowly fails to 0.009
that is over around 20sec.
..or this meter is pure crap?