So close and yet so far.... The "Heavy" preamp mess continues

The TL783 data sheet shows that for an output voltage of 50V it has a ripple reduction af about 60dB so you would expect your 1V at the input cap to be reduced to 1mV. But you see 20mV so something is definitely no right. With 75 ohms between output and adj pins you will be drawing 15mA minimum. Your phantom powered mic load is probably less than a third of this so it is not surprising the ripple does not change much. 20mW ripple on a phantom supply is definitely way too much. In my version using the TL783 I have a 10uF capacitor from the adjust pin to 0V to improve ripple rejection. The data sheet does no recommend doing this because it can affect transient load response but we are not really worried about that.

Your tripler is half wave so the caps have to work twice as hard and because there are three stages they are one third as effective at reducing ripple. Increasing them all to 200uF would not hurt. Just increasing the last one is not enough.

Cheers

Ian

ruffrecords said:

The TL783 data sheet shows that for an output voltage of 50V it has a ripple reduction af about 60dB so you would expect your 1V at the input cap to be reduced to 1mV. But you see 20mV so something is definitely no right. With 75 ohms between output and adj pins you will be drawing 15mA minimum. Your phantom powered mic load is probably less than a third of this so it is not surprising the ripple does not change much. 20mW ripple on a phantom supply is definitely way too much. In my version using the TL783 I have a 10uF capacitor from the adjust pin to 0V to improve ripple rejection. The data sheet does no recommend doing this because it can affect transient load response but we are not really worried about that.

Your tripler is half wave so the caps have to work twice as hard and because there are three stages they are one third as effective at reducing ripple. Increasing them all to 200uF would not hurt. Just increasing the last one is not enough.

Cheers

Ian

Click to expand...

Good to know.  On Monday I'll be better able to see what parts I can scare up without having to order anything.  Certainly willing to swap out some caps, because it's the cheapest, easiest option at this point.

Once I get this licked, I can't see what else pops up in this little game of electronic whack-a-mole!

Made a little progress this morning.

Replaced ALL of the capacitors in the phantom supply with 220uF/100V.

Fired up the supply, and got the correct voltage.  So far so good.
Checked for ripple unloaded and didn't see any.  Since I'm working this morning from a different bench, I only had the power supply with me, so to test it loaded, I rigged up an XLR jack with a couple 6.8K resistors and hooked up the supply.  Then I plugged in a condenser mic to test it loaded.  No audio, of course, but we're just looking for ripple.

Couldn't find any.  Had to turn the scope down to about 10mV/div to see a little wiggle in the trace, but nothing even close to that nasty square wave mess I was seeing before.

Obviously, I'll reconnect it to the audio circuit on my own bench tonight or tomorrow and retest everything the right way, but for now it looks like I'm making progress.

SUCCESS!!!

So far.....  ;D

Connected the new and improved power supply to the audio circuit and fired it up.  NO MORE HUM!!

Well, I shouldn't say "NO more," but  a whole lot less.  What hum I heard was at extreme gain settings, and I'm not sure how much of it was actually coming from my headphone amp.  (I really should find a quieter one for work like this...)  Also bear in mind that the thing is still basically a pile of guts on the workbench with no metal chassis to speak of around it, and the tubes are not shielded.

I did run into a slight problem when using a dynamic mic.  If the gain was turned up past about 2:30, I was greeting with a very high pitched tone.  Ouch!  Now what??  Turns out the remedy was simple.  The small length of balanced cable (about 3-4 inches) between the input XLR and the input transformer only had its shield connected at the transformer end.  Using a small clip lead, I connected the shield at the connector end to Pin 1, and the SQUEEEEE!!! was gone.  The condenser mic worked well.

That's all for tonight.  Pretty likely going to be snowed in tomorrow, so I might have time to clean up the wiring somewhat and maybe hook up the other channel.  If that goes well, I'll start adding features back in.

CurtZHP said:

SUCCESS!!!

Click to expand...

Whoopee!

Ian

ruffrecords said:

Whoopee!

Ian

Click to expand...

I know, right?!

So, back to the bench....

I thought I'd have another look at the other two rails on this supply (12VDC and B+), to confirm what sort of ripple was going on there.  The heater supply was showing a little at 10mV, much "cleaner" than what I'd seen previously on the phantom supply.  The B+ showed a little between 5mV and 10mV.  Are these anything to be concerned about?

Shifting gears, as I study the topology of the audio section...
I've noticed on some designs they put a "grid stopper" resistor in series between the input and the grid (not to be confused with the 1Meg resistor going to ground...).  I don't have one.  Most of what I've read suggests that this helps with RF, so I suppose I should add one.  Thoughts?

10mV ripple on the heaters is nothing to worry about. 5 to 10mV on the HT is too much for a mic pre. Looking at the first stage of the pre wou have a 12AX7 with a 100K plate load. The power supply rejection ratio (PSRR) of this stage is determined by the ratio of the tube's plate resistance and the plate load. Let's be generous and assume the 12AXy ra is only 50K; this gives a PSRR of just 10dB or about 3 times so your 5mV ripple will appear as about 1mV at the plate of the 12AX7. Not what you want.

You could add some extra power supply decoupling to the first stage. Adding a 10K series resistor and a 10uF capacitor to ground will drop the ripple by more than 60 times to a few tens of microvolts. Even so, this is a bit high for my liking so it might be worth paying some more attention to the HT supply. One problem you have is the resistor values setting the output voltage are much higher than the data sheet recommends. I am not sure how this affects ripple reduction. As I mentioned a while back, you need a good 20V across the TL783 for it to regulate and ripple reduce properly. It looks like you have your HT set at about 250V. With 22VAC going into the bridge you should get over 300V across the reservoir so that looks OK.

I don't use regulators for HT. Tubes don't really need a regulated supply but mic pres need a low ripple supply. I use three stages of RC smoothing in most of my designs. This is a lot better than a single stage of 3R and 3C. I typically use 1K and 220uF times three. Each stage gets you over 40dB of ripple reduction so the three together get you over 120dB. The TL783 struggles to reach 80dB. 120dB of reduction will drop a 1V ripple to 1uV.

Grid stoppers are not really necessary for low gm tubes like the 12AX7 and 12AU7. Their principle use is to stop high mu tubes hooting at VHF. Adding one to the first stage grid only worsens the noise.

Cheers

Ian

I went back over the thread, and over my notes, and discovered a couple changes that Jakob had recommended, which I forgot to implement.  I redid the schematic to reflect them.  Do you suppose such changes would help with ripple on the HT supply?

(I think they were more intended to provide extra protection to the regulator...)

The resistor values setting the output were plagiarized from his G9.  Following the spec sheet, I figured on updated values, but I'll have to dig for those.

I do not see anything there that would materially affect the ripple. The zeners are there to protect the TL783.

Cheers

Ian

ruffrecords said:

I do not see anything there that would materially affect the ripple. The zeners are there to protect the TL783.

Cheers

Ian

Click to expand...

Yep.  That's what I thought.  So, hunting for some better suited output resistors for the 783.

Meanwhile, you mentioned better decoupling of the power supply for the first stage.  I assume that would happen upstream of the plate resistor?

Here's an updated schematic reflecting what I think would do what you suggested.  Look at the plate supply for the 12ax7.

I happen to have a 15uF/350V cap on hand.  Would that work as well as a 10uF?

Yes, 15uF will be fine.

Cheers

Ian

ruffrecords said:

I don't use regulators for HT. Tubes don't really need a regulated supply but mic pres need a low ripple supply. I use three stages of RC smoothing in most of my designs. This is a lot better than a single stage of 3R and 3C. I typically use 1K and 220uF times three. Each stage gets you over 40dB of ripple reduction so the three together get you over 120dB. The TL783 struggles to reach 80dB. 120dB of reduction will drop a 1V ripple to 1uV.

Ian

Click to expand...

Sorry for intruding the thread but I became curious about those 220uF capacitors? How big are they? What voltages are they rated for? I am using 1U chassis and 100uF/400V are the biggest I can find to fit inside a 1U chassi (radial capacitors that is).

Best regards

/John

ruffrecords said:

Yes, 15uF will be fine.

Cheers

Ian

Click to expand...

Tried it with the 15uF and a 2.2K resistor (couldn't find a 10K, and I noticed that Jakob's used a 1K...) and it actually made the ripple WORSE!  A lot worse!  Like 200mV!

Either I did it wrong, or the cap is no good.  In any case, something's wrong.

johnheath said:

Sorry for intruding the thread but I became curious about those 220uF capacitors? How big are they? What voltages are they rated for? I am using 1U chassis and 100uF/400V are the biggest I can find to fit inside a 1U chassi (radial capacitors that is).

Best regards

/John

Click to expand...

The 220uF/350V caps I used are 2 inches tall and 3/4-inch diameter, so to put them in a 1U chassis you'd have to lay them on their sides.

CurtZHP said:

The 220uF/350V caps I used are 2 inches tall and 3/4-inch diameter, so to put them in a 1U chassis you'd have to lay them on their sides.

Click to expand...

You can get short fat ones such as these - only 30mm (1.2 inches) tall:

http://uk.farnell.com/epcos/b43644c9227m000/cap-alu-elec-220uf-400v-snap-in/dp/2468318?st=220uF

Cheers

Ian

Question about testing for ripple on the plate supply....

Is it possible, or even advisable, to test with the first gain stage actually connected to the supply?
What I'm concerned about is that I'm using the scope to essentially measure AC on a rail that happens to be carrying audio as well as supply voltage, right?  So, if the input stage is present, that's going to introduce whatever noise is present in that stage (like cable noise, mic output, etc...), skewing whatever I see on the scope.

What, you don't have a Filter Cap there?

WHY are you posting what amounts to an Internet Poll? The ignorant, idiotic, and off-point answers always "win".

Just do it. Ripple looks different from cable-twitch or hiss.

PRR said:

What, you don't have a Filter Cap there?

Click to expand...

Not yet.  Gimme a minute....

PRR said:

WHY are you posting what amounts to an Internet Poll?

Click to expand...

'cause I'm just not as smart as you, quite honestly.

CurtZHP said:

Question about testing for ripple on the plate supply....

Is it possible, or even advisable, to test with the first gain stage actually connected to the supply?

Click to expand...

Yes

What I'm concerned about is that I'm using the scope to essentially measure AC on a rail that happens to be carrying audio as well as supply voltage, right?  So, if the input stage is present, that's going to introduce whatever noise is present in that stage (like cable noise, mic output, etc...), skewing whatever I see on the scope.

Click to expand...

You test it with no audio. Short the input to minimise noise. You are just looking at the twice mains frequency ripply. It will be a lot higher than any noise created by the mic pre itself.

I don't risk my scope by connecting it directly to a big dc voltage with a small ac signal on it. I use a 100nF/400V cap from the HT supply to a 100K resistor the other side of which is connected to ground and measure across the 100K

Cheers

Ian