Maddening 100Hz ripple hum

[Alexandru marian]

I am going slightly crazy over a bit of 100Hz hum (ripple, not 50Hz harmonic; there are no 50Hz transformer etc problems) I have in my G9 mic preamp.  Crazy not because I can hear it when recording, but because it seems to be random/intermittent. 

The HT psu is a CR regulator RC. Originally the two caps were 200uF and the hum was pinned to about -65db on medium gain when acting up and down to  -75 when doing fine. 

I then bought a 400uf poly(!) for the second cap and bundled the old ones together for the first.  Hum went down to -78 when acting up, or to about -90 (pretty much at general tube noise level on all but the lowest gain)

What drives me mad is why is it intermittent.  At all times my DMM shows 0.000 AC on both caps.  The grounding seems sound. I've tried a few different ground schemes and mods with no effect. 

For a couple days I left the amp under constant 1K sine and looked from time to time at the 100Hz bump. It can be there for a few hours then it goes away for a while then back again.  There is nothing in the house that can be associated with the change.  There are times when it looks good and everything else draws current: computer nearby, lights, fridge, TVs everything. 

So i am suspecting something odd coming from the mains. Or something fishy within the regulator.  Or perhaps this is just about the best this particular reservoir size can do.  Can't really up it any more for space reasons, the 400uf poly is the size of a milk bottle. 

Would adding a small value resistor in between the parallel first two caps improve it a bit? 
Again, at -78db I can't hear it in recordings but I can't get over it. The overall noise level of the amp is excellent and pretty much on par with 3 other solid state pres I have. 

Thanks!!

 

[Alexandru marian]

This is how it looks when it is OK, at roughly 20, 40 and 55db of gain. 

 

[Alexandru marian]

This is how it looks when it acts up. Today seems to be a real bad day, up to -72 on the highest gain: 

 

[Alexandru marian]

HT suply currently looks like this.  What if I move one of the 200uF caps before the first resistor? 

 

[JohnRoberts]

What are voltages on input and output side of the regulator? 
What are the voltages of those clamp zeners?

If unregulated voltage is high enough for the zeners to conduct that could introduce ripple into the regulated output V.

JR

 

[Alexandru marian]

Thanks.  The Zeners are 39V.    Voltages are like this:

mains 230, rectified 390, first resistor / cap  380.    After first zener 336, second 292, third 249.  Regulator heatsink 249, after second  resistor / second cap 240. 

 

[ruffrecords]

You should have a cap from the ADJ pin of the TL783 to 0V and also one directly from the out pin to 0V for best ripple reduction and stability. Check out my phantom power supply design which uses the same device. Look in the Power folder of my web site:

http://www.customtubeconsoles.com/diy

You should also add a spike suppression capacitor across the ac input unless you are using very fast rectifier diodes.

Cheers

Ian

Edit: The TL873 also needs a minimum load current in order to properly regulate. This is normally achieved by the bias network that sets the output voltage but in this case the resistor values are too high to do this. Does another part of the ciruit provide this minimum load?

Cheers

ian

 

[JohnRoberts]

Thanks.  The Zeners are 39V.    Voltages are like this:

mains 230, rectified 390, first resistor / cap  380.    After first zener 336, second 292, third 249.  Regulator heatsink 249, after second  resistor / second cap 240.

By my math 3x 39V zeners adds to 117V, 
380V -250V= 130V

So 117V of zener diodes with 130V across them "should" be conducting current.

The output voltage should be clean (relatively) DC, the input voltage will have ripple that may make the zener current even worse at ripple peaks.

I don't want to redesign a product I don't even know what it does, but perhaps increase the value of the 470 ohm resistor to scrub off some voltage (may need higher wattage resistor).  From another perspective what exactly is the function of those zener diodes (i will speculate to protect the regulator from over voltage but I am too lazy to look up the data sheet)?

The math seems off somewhere for that design, maybe too much transformer. Try a larger resistor for the 470 ohm?

JR

 

[ruffrecords]

Looks to me as though the TL783 is not regulating. It has a maximum withstand voltage of 125V between input and output pins. The zeners I guess are intended to enforce this under fault conditions. However, given the measured voltages it looks like the zeners are operating under 'normal' conditions. I suspect your TL783 is dead.

Cheers

Ian

 

[Alexandru marian]

Ian:

many thanks on the additional caps tip.    Luckily I had a couple 6.8uF 250V polys and I added them from adj and out.  Being relatively small I didn't expect anything but they provided a solid 10db reduction!  I have the feeling this is not the random "nice phase", as it is still not as a low as I had measured yesterday, but close.  Today the mains seems "bad" at all times, yesterday it was intermittent.  If this is anyway related to the mains...  But will keep an eye on it. 

The rectifier diodes are plain 1N4007,  will change to something fancy one of these days. 

Regarding the TL regulating, I think it kicks in when it sees load from the tubes. Seems to work: when I turn on the amp the high DC goes right through it.  Final reservoir has over 300 while tubes heat up then it goes down to 240. 


John,

Yes I think the diodes are protection.  It might very well be too much input.  The original schematic called for 15 to 12V back to back transformers.  That should give 15/12x230=287V so the 275 step-up I bought (the back to back I originally had gave heat and mechanical buzz) should have been close enough.  For some reason in practice, the AC before the rectifier diodes reads 310.    I don't know how this turns to 390DC.  With a 4 diode bridge one would multiply with 1.41

 

[Alexandru marian]

Ian,

I have spare regulators - but if it is dead, how come I still get the proper voltage?  When powering up, it first passes the full DC but then settles to 250 after less than 1 minute. 

And if it is dead, what could have killed it?  I don't recall ever shorting the HT.  I did blow a diode (from the heater bridge I think) once when I tried to feed the heater and phantom supplies in parallel from the same transformer winding but I don't see the relation to the HT.

 

[JohnRoberts]

First things first... the unregulated voltage looks too high... increase 470 ohm

However if that voltage drops under load, maybe we need to measure it under load.

When are you getting the hum you are worried about?

JR

 

[Alexandru marian]

It is under load - power on with signal passing through. 

The hum appears at most times. Sometimes it goes away but most of the times it is there.  I can't hear it on a recording, it is low enough but it bothers me on a "theoretical" level. 

Ian's ADJ cap trick removes most of the ripple but I need to try with lower voltage too, so I'll next add an additional resistor between the first two 200uF caps and replace rectifiers with MUR1100.  Will get back in a few days after I receive the parts. 

 

[Matador]

I think Ian is right, as I've been bit by the same thing.

The TL783 needs 15mA minimum load current to maintain regulation.  I would bet that the 'bad' days are ones where you are using a low(er) current microphone which isn't loading the phantom supply enough.

 

[Alexandru marian]

The load from the 4 ECC82s as stated by Jakob is 20mA and this matches perfectly the 10V drops across the 470 resistors.

While I wait for a batch of new resistors, I thought about moving the last one in front of the regulator.  I also changed the first filter order out of curiosity, so it now starts with a cap rather than resistor. 

I do not know why but the voltage is even higher now, the DC starts at 410 instead of 390.  The difference across the regulator is still above 125V, 132V now, used to be 130.  The good news is that the 100Hz hum is super low.  What is a bit worse now is the 50-150-250 series on the lowest gain. 

 

[Alexandru marian]

On a totally different note, I wish I'd knew what is this strong level from 0 to 15Hz. 

 

[JohnRoberts]

The load from the 4 ECC82s as stated by Jakob is 20mA and this matches perfectly the 10V drops across the 470 resistors.

While I wait for a batch of new resistors, I thought about moving the last one in front of the regulator.  I also changed the first filter order out of curiosity, so it now starts with a cap rather than resistor. 

I do not know why but the voltage is even higher now, the DC starts at 410 instead of 390.

of course ... that is the "opposite" of my advice to increase 470 ohm to reduce unregulated voltage. Now there is 0 ohm between diodes and cap so the reservoir cap charges with full transformer current to even higher peak voltage.  (more resistance there will therefore deliver LESS unregulated voltage).

  The difference across the regulator is still above 125V, 132V now, used to be 130. 

do you understand that the zener clamps are only supposed to conduct during start up or to protect against faults? Not all the time.

The good news is that the 100Hz hum is super low.  What is a bit worse now is the 50-150-250 series on the lowest gain.

I hate repeating myself but I am used to it from discussing politics around here.  :

Increase 470 ohm in series with transformer to reduce unregulated voltage so zener diodes are not conducting current around the regulator and corrupting that DC regulated output voltage during normal operation.

250V at regulator output, + 3x39V = 367V. To be safe and keep the zeners turned off I would aim for around 350V unregulated...

JR

 

[Matador]

Oops disregard what I said I thought this was the phantom supply.

JR is right, you need to make sure those Zeners are shut off except under 'non-normal' conditions.  They are simply there to limit the differential across the regulator when the B+ supply charges from 0V (or is accidentally shorted).  Once the Zener's turn on, they will 'override' the regulator and pull the output voltage up with them.  Since this happens periodically with the input cycle it will appear as 100Hz hum.

The 'good' days may just be when your incoming line voltage is lower by a few volts.  You are operating very close to the Zener turn-on voltage so your results will change from day-to-day.

 

[Alexandru marian]

John, please don't shoot me!  (and thanks for still trying to help)   

After all I did increase the total resistance in front of the reg.  Putting the cap first was out of curiosity - more often than not PSU schematics seem to start with a cap. 

Will flip them now. 

Matador, mains varied in between 228 and 232 over the past few days.  Could be it, but the AC out the transformer looks rather stuck at 306.  Btw it is rated 275!  Perhaps the load is too small for it? 

 

[JohnRoberts]

John, please don't shoot me!  (and thanks for still trying to help)   

Sorry but I get frustrated..(I'm old and a little cranky). I will try to explain more fully...

After all I did increase the total resistance in front of the reg.

But that is not literally what I asked for.

Adding a resistor after the smoothing capacitor will drop voltage based on the average DC current (20mA?)

A resistor before the smoothing cap limits the cap charging current that only flows for the small part of the waveform that winding voltage is higher than the sagging capacitor.  Since to charge the cap to supply 20 mA average, will require several times that current for the shorter charging duration. This several times higher than average current means the extra 470  ohm resistor will drop several times the 10V it drops from 20 mA.

The math for this is not trivial so please trust me, the same value resistor will scrub off more unregulated voltage  when placed before the reservoir cap than after.   

Putting the cap first was out of curiosity - more often than not PSU schematics seem to start with a cap. 

Yes if you are trying to get the most voltage from the transformer, but you have too much voltage.

Will flip them now. 

Matador, mains varied in between 228 and 232 over the past few days.  Could be it, but the AC out the transformer looks rather stuck at 306.  Btw it is rated 275!  Perhaps the load is too small for it?

I suggested in a post yesterday that it may be too much transformer, but adding/increasing  a resistor value is cheaper/easier than buying a new transformer.

  JR