Maddening 100Hz ripple hum

[Alexandru marian]

Yesterday i rewired what I have to 470R 200u 470R 200u Reg 400u - of course not enough but anyway resulting in 306VAC, VDC: 390 379 369, then 327 288 249 across zeners.  Today I notice mains is low at 226.  Check inside:  302 VAC, 384 VDC 375 366 - so far so good, reasonably in line with the mains drop.  Then zener drops: 320 271 225! 

 

[JohnRoberts]

Yesterday i rewired what I have to 470R 200u 470R 200u Reg 400u - of course not enough but anyway resulting in 306VAC, VDC: 390 379 369, then 327 288 249 across zeners.  Today I notice mains is low at 226.  Check inside:  302 VAC, 384 VDC 375 366 - so far so good, reasonably in line with the mains drop.  Then zener drops: 320 271 225!
[/quote
Isn't the regulator output 250V?

unsure about 320-271-225 zener drops... regulator should deliver 250V out from 320V input?  Is it possible regulator is faulty.

JR

 

[Alexandru marian]

Based on the (simplified) formula, it is set at aprox 243V.  I think it is toast too, but I won't put a new one until tomorrow when I get the new resistors.  Currently the noise is very low. 

Btw, the first caps used to drain very fast after powering off the amp. Today they stay stuck at 120V even after the last cap is empty.  I don't know what was draining them, the regulator or the final 470R (which is now removed) but in any case now they do not drain. 

With new parts first test will be :      rectifier  2K    470uF (new single cap)    new reg    400uF poly.     

 

[radardoug]

Those cap values are ridiculously high. Normal cap values for high voltage circuits are 22 to 47 uF. Why are you using such high values? Somewhere in there you should have a bleeder resistor which will discharge the caps when the supply is turned off, typically 100K to 470K.

 

[ruffrecords]

Those cap values are ridiculously high. Normal cap values for high voltage circuits are 22 to 47 uF. Why are you using such high values?

Historically values in the region of 47uF were used because at the time they could not economically make higher value high voltage electrolytics. If higher values had been available they probably would have used them. Today 470uF 450V caps are common and cheap because they are used is SMPSUs. A tube mic preamp needs a particularly low HT ripple and one of the simplest ways of achieving this is using several RC stages. The ancients realised that cascaded RC stages produced better ripple reduction that a single Rc of the same total R and C:

https://drive.google.com/file/d/0B_n67A1hN3qtUWJDRFlnenlZODg/view?usp=sharing

It is quite easy to obtain better than 90dB of ripple reduction with three RC stages, better than any regulator and it's a lot more robust than a regulator.

Cheers

Ian

 

[radardoug]

Yes but he's using two stages and a regulator, should work fine, although running with such high input voltage means the zeners are just transferring the ripple.

 

[ruffrecords]

Yes but he's using two stages and a regulator, should work fine, although running with such high input voltage means the zeners are just transferring the ripple.

I agree.

Cheers

Ian

 

[Alexandru marian]

I have the impression a small cap works better with a choke. Weren't those pretty much standard in the old days? 

Back here, the regulator was definitively dead.  Still no new parts in but, I remembered my cap sucker is a 5K 2W.  Just perfect, I put it in in place of the regulator for a HT of 260-265.  So it is now simply:    rectifier  470R  200u  470R  200u 5K  400u

Noise looks very good and that funny rumble under 10Hz is reduced too.   

I will still try with a new regulator, curious if it clears those 50-150-250 spikes on the lowest gain although they are probably magnetically induced by the transformer.

 

[PRR]

> regulator was definitively dead

What I was hinting in your parallel thread on the other forum.

390V - 250V is well past the regulator's limits. Protection diodes may or may not save it; if they do, it is by BYPASSING the regulator (so why have it?).

Don't put up with "maddening" tricks. Especially when there are tricks more suited to ~~400V use, and you DON'T actually need strict voltage regulation.

Since this is for a single known stable load, Ian's complaint that the Ancient Engineers have stolen his discovery is on-point. Multiple R-C filter IS good audio.

> 470R  200u  470R  200u 5K  400u

Yeah. Elsewhere I computed like 470-200u-2.5K-200u-2.5K-200u-2.5K-200u which will give some more 50/100hz filtering and a lot more >200Hz filtering for your cap budget (as Ian's file shows in antique language). But even 5K-400uFd is heavy filtering.

 

[Alexandru marian]

I can't split the final 400u, and new cap I bought also is a single 470.  Idea was to add up the old 200 caps to my microphone psu which needs more filtering too.    But I can get extra parts and make a long RCRCRC if the new reg fails again.  I also have an itch towards trying a choke

 

[JohnRoberts]

Unless the choke (a coil) is magnetically shielded it might pick up noise from the transformer's magnetic field (depending on orientation).

There is nothing wrong with KISS.  The LF noise may be 1/F noise from the regulator. It my last day job I black balled one 3 terminal regulator maker because of noise that the MFR refused to even specify.

JR

 

[ruffrecords]

Noise looks very good and that funny rumble under 10Hz is reduced too.   

I will still try with a new regulator, curious if it clears those 50-150-250 spikes on the lowest gain although they are probably magnetically induced by the transformer.

With this sort of thing, experimentation is a good way to learn. Often it is surprising what you need to do and equally often what you don't need to do.

Cheers

Ian

 

[Alexandru marian]

OK, new psu tested. 

mains 227VAC (lowish)  306 from transformer, MUR1100E diodes, 2K 5W Mills resistor,  470uF,  new regulator, 400uF. 

It finally works, there is no big voltage across the zeners (251 and 256V) and output is good at 252V.      The 2K drops about 40V as hoped for. 

But the DC after rectification is only 336!  Used to be 390 (410 if capacitor first).    This should be because of extra load from the working regulator?

As for the noise, well I got back the 100Hz spike.  It is not horrible but it is there.    If it will go up as the mains increases beyond 230V (it should not with the reg working)  I'll maybe put back an ADJ cap or just move to RCRC(RC)  Or maybe not.  Just did a recording and it sounds very good.  The 100 spikes was driving me nuts because I didn't know what caused it and why was it intermittent.

50/150 etc spikes at low gain still there.  Now I suspect they are not from the transformer (which happens to have a mu shield) but from the mains wires going to the front power switch and back. 

 

[Alexandru marian]

Lesson of the day is that you can't count on tl783 to act as the R in an RC!

I changed it to:

Rectifier 1K cap TL@275V  1K cap 257ht

100hz is completely gone now, 100% gone even at the lowest gain.

Now let's kill the 150,250,350 etc

 

[ruffrecords]

Now let's kill the 150,250,350 etc

I think I mentioned before that higher harmonics can be caused by spikes as the rectifier diodes turn of. The usual cure is a 100nF class X 275VAC rated fil cap across the HT transformer secondary winding near the rectifiers or one across each rectifier diode.

Cheers

Ian

 

[Alexandru marian]

These diode spikes infiltrate through the air or through the wire? I did the tests on the chanell further away from the psu and no spikes at all. 

I elimated the mains wiring to fron switch, no change at all.  So it can be from the toroids.  Maybe not the main one as it has a mu shield but the separate 48v unit.

I started to drool over nice traditional transformers, for example lundahl 1683.  includes phantom tap and it is a lot more compact than the big toroid - enough space to put a 1cm thick shield if needed :-D

 

[ruffrecords]

These diode spikes infiltrate through the air or through the wire? I did the tests on the chanell further away from the psu and no spikes at all. 

Which implies they travel through the air. If you stop them at source it does not matter how they travel.

Cheers

Ian

 

[PRR]

> 306 from transformer, ...diodes, 2K 5W Mills resistor, 470uF
> But the DC after rectification is only 336!  Used to be 390 (410 if capacitor first).

You have tried several different values for the resistor between diodes and first capacitor.

Of course you will get different DC voltages at that first cap.

 

[Alexandru marian]

I was referring to the dc right after the diodes, not yet chiped by R.  i guess the transformer load varies with what the R are chewing not just the tubes.

 

[JohnRoberts]

I was referring to the dc right after the diodes, not yet chiped by R.  i guess the transformer load varies with what the R are chewing not just the tubes.

Don't know what chiped means but the voltage right after the diodes is not a simple DC or AC voltage.  For parts of the AC waveform higher then the following capacitor voltage it will roughly follow the rectified sine wave, for parts of the AC waveform lower than the capacitors DC voltage the diodes will not conduct so the meter will read the DC voltage of the capacitor.

JR