Valve Microphone PSU noise

I recently bought a pair of Hammond M100 nuvistor condenser mics.

They can be seen here.


The PSU is strange. It is a valve / transistor hybrid. The HT section uses a 0C2 valve voltage regulator! The heater section is solid-state.

Anyway, the audio output of the PSU is unbalanced, but there seems to be quite a bit of noise despite moving the cable around (I will be balancing the output with transformers). I scoped the voltages going to the mic which are -5V for the heater and also ~115V and 90V (not sure why two HT voltages). Anyway, I was more interested in the PSU noise. There is 100mV of 100Hz noise on each of the power supply pins. Is this too much? The caps are likely quite dry which is why I thought I better check for ripple.

Cheers,

Roddy

Looks like fun!

I'd recap the power supply and any other electros just as routine maintenance, and be generous too with caps in the PSU - you can usually fit in substantially larger values with modern caps.

Not sure about the two voltages. I guess it will make sense when we see the schematic. :wink: Perhaps one is polarisation?

I haven't worked with nuvisters so I don't know their little idiosyncracies!

Good luck.

Thanks, Stewart.

I actually have some good news; the Neutrik audio test set which I used as a mic amp at work didn't seem to be working properly. I've just used it with a mic preamp at home and there is no hum at all.

Both mics sound nice. Big low-end extension and a nice bright top-end. But, one of the mic's has louder background noise than the other. It's a sort of constant white noise, but with less treble frequencies. Sounds like a constant wind blowing. The circuitry looks a little dirty (possibly covered in oil/WD40). Should I give it a clean with meths? Or maybe the capsule is moist.

Roddy

Excellent news!

Clean is good. A good grade of IPA / isopropyl alcohol / propan-2-ol is better. Meths generally contains a lot of crap to stop people drinking it.

You could try popping the mic / capsule in a bag with some silica gel. (Dry the silica in an electric oven an lowest setting, or zap it in a microwave beforehand).

(I'm still having noise trouble with my Teladi, but more of that later.)

I've just swapped the 400uF output coupling caps for Elna long-life versions which should be happier next to the valve voltage regulator (WTF!) The old caps were reading 190uF.

I will spray it with IPA tomorrow. I actually used it at work when cleaning the supply and managed to skoosh it into my eye. Bloody sore. COSH? Meh.

I'll try the silica gel trick tonight too.

Working on a schematic thsi evening (it's very simple) but I need to get a pin-out for the nuvistor used. Also, I can't get the nuvistor number just now as it would require serious dismatling.

I've now realised each mic is as noisy as the other. I'll try the silica gel, and will spray some IPA (may drink some too) although the resistor values are 82 M, so resistances bridges from dirt, etc are going to be less of a problem.

Here is a schematic:



And here is a picture of the capsule:

Is the cathode resistor 130 ohms?

What is the voltage at the cathode?

I am wondering what the current is in the circuit?

Yes, the cathode resistor is 130 Ohms. I know that seems quite low.

The bias voltage is 0.76V which as expected is quite low, although the mic has a relatively low output for a condenser, so maybe it can keep the output swing within range.

Why are you interested in the current flow if you don't mind me asking?

.76VDC /130 ohm about 5.8ma

And the tube is powered by 115VDC at the plate

I am trying to understand the design

What I mean AKG had CF microphones that got the capsule voltage from the bias and cathode resistors only two supplies B+ and fil
http://www.geocities.com/SiliconValley/Peaks/5212/start.htm
c60 c61

5.8ma what input resistance were they designed to drive? why 115VDC?

The link to the Revox has 1 meg input resistance in the specs

[quote author="Gus"]
I am trying to understand the design

What I mean AKG had CF microphones that got the capsule voltage from the bias and cathode resistors only two supplies B+ and fil
http://www.geocities.com/SiliconValley/Peaks/5212/start.htm
c60 c61[/quote]

Cool. I was surprised myself that there was a separate polarisation voltage. Saves a little space in the mic though (they are very small).

5.8ma what input resistance were they designed to drive?

Click to expand...

I'm not sure. They are unbalanced, but the output impedance is very low (70 Ohms or thereabouts). I don't see why they can't drive 2K without any problems.

why 115VDC?

Click to expand...

I'm guessing this is tied to the use of a valve regulator. Regulation was easier for filament voltages using germanium, but this isn't good for B+ voltages. The 0C2 valve regulator chosen seems to work at this voltage of 115V from what I gather.

The link to the Revox has 1 meg input resistance in the specs

Click to expand...

1 Meg? Seems awfullt high. The Beyer-made, Revox-branded mics were increased from 200 Ohms to 600 Ohms, but having a 1 Meg input seems a bit pointless to me.

Considering the operation of the nuvistor, having a 2K mic input impedance sitting in parallel with the cathode resistor is unlikely to upset the operation as far as I can see.

> I can't get the nuvistor number

Almost has to be 6CW4 or its industrial twin. And, as I see you guessed, the H-H-K-G-P pins are the same on all Nuvistors.

For 115V supply and 130 ohm Rk, it will bias near 7mA, Gm ~~9mS or 110 ohms, effective output impedance of 110||130 is 59 ohms, just as the ad says. THD will be non-zero (but sweet) at 0.5V peak, and it probably never gets there.

7586 will give a bit more headroom by sucking twice the current, but otherwise will work the same.

If we assume noise is all in the 1/Gm cathode resistance, hmmmm... resistances at different temperatures... I guess the noise is like a 100 ohm resistor at room temp, or like a dynamic mike. Actual noise will be higher due to grid current, grid resistors, capsule impedance, and acoustic Brownian noise in the capsule.

The cable has +90V and +115V because the capsule needs better filtering than the CF, and there is no room in the mike for a filter, so it is in the box.

You could do things different. I don't know what you could do "better", unless you work with absurd high levels, or very harsh environments.

While Nuvistors usually live forever, this one is running right at rated dissipation. I'd still presume "good", because it probably has just 10 or 20 hours of use.

I'd think you could "dry out" water moisture by leaving them powered overnight. They must run near hot enough to burn your fingers. Other types of contamination will not come off so easy. (I just tried to solder some leads which had sat under a tank of hot silicone oil for a decade...)

The 5V heater is lower than spec, but may be a compromise with noise or skin-burns.

> 5.8ma

Yeah, well, 7mA, 5.8mA, whatever.

All tubewerk is +/-20% anyway.

And it could be related to heater 6.3V nominal but 5V on rodabod's sketch. They dropped cathode temp until current and Gm fell 10%-20%?

> why 115VDC?

I'm not sure why it isn't 90V. But the general goal would be HIGH current for high Gm and low output impedance without a transformer. 5786 could get big current at 90V, but maybe 6CW4 was soooo much more available. 115V 7mA is jammed up in the top-right of the allowed dissipation curves... as hot as possible favoring high Gm.

> what input resistance were they designed to drive?
> The link to the Revox has 1 meg input resistance in the specs

The Revox input is "naked grid" so it can be used with high impedance mikes, with whatever transformer, or plausibly with crystal/piezo mikes (which were still trying to nibble-out a corner of Hi-Fi).

The mike output is designed to drive long cables without treble loss, and with some ability to absorb buzz and radio on its unbalanced line. I run higher-Z unbalanced lines hundreds of feet. Unless you drape it over the elevator/lift motor and under the AM transmitter tower, it will be clean. The low Z also comes out of low noise low-cost design.

The "mismatch" into the Revox 736 is unimportant. The mike don't care. If all mike noise were tube Gm, then using it straight into the 736 raw grids gives up a few dB of noise figure, but perhaps no worse than a dynamic. We all ran 200 ohm dynamics into grids and bases without input iron and got away with it. 'Specially at affordable tape speeds.

It should equally well drive modern "Pro" XLR inputs, assuming the power-box takes off ALL the DC before your Pro transformer.

Looks like you got yourself some nice little mics!

What's in the PSU? Is there an output cap somewhere? Must be. What's the value? Is it an electro?

[quote author="zebra50"]What's in the PSU? [/quote]

A whole bunch a shit! Well, the usual suspects plus some more strange additions. There's a transformer, a diode bridge, some large electros, and then there's a germanium OC*** power transistor (I assume part of the heater supply section) and there is also the 0C2 valve voltage regulator which makes an awesome plasma glow.

Is there an output cap somewhere? Must be. What's the value? Is it an electro?

Click to expand...

Yeah, the signal from the mic is coupled to the output via a 400uF electro. I measured these as 200uF so have swapped them with 470uF Elna long-lifes as they sit next to the valve regulator.

I've just measured some of the 82M resistors which I reckon are in the range of 90M. Should I be worried that these carbon comp resistors might have gone noisy?

I'll replace the PSU caps as you suggested Stewart, but I'm wondering how large I should go in case I get into problems with cap inductance or changing RC filter values. I'm going to trace a little of the PSU just now.

Cheers,

Roddy

Ok, here is a MS Paint drawing of the high voltage section of the power supply.

As you can see, each filter stage as added the the previous in a sort of ladder fashion. The 90V polarisation voltage is tapped off early, and the 115V B+ is taken from the resistor before the voltage regulator. I think this 1.2K resistor is to bring the voltage a little higher than the standard voltage you would see across the 0B2 regulator.

What I'm wondering is, would it be sensible to stick to the current values of 33uF in order to maintain the filter frequency points of each of the stages, or should I raise the values of these caps? Or alternatively, should I keep the values but replace the first 33uF with a larger value since it doesn't have a series resistor? I noticed Jakob used 2x 470uF caps for the HT section of the G7. Also, perhaps I could add local decoupling inside the mic for the polarisation voltage if I can fit it.

The heater section is filtered with four 1000uF caps which seems reasonable to me. I wonder if an additional smaller cap may be beneficial though.

Hi Roddy,

Something looks not quite right - 95V is lower than 115V but appears closer to the diode bridge. I know there's a 10Meg, but this doesn't appear to be part of a voltage divider. What's dropping the voltage? Either there is a resistor missing, or the two voltages are backwards, or...?

Nice use of the OB2 voltage regulator to higher than 108V!

I would just look in my spares box and sub in some bigger caps according to what fits. It certainly won't do any harm. Some people do add a small poly cap in parallel with the electros. They are allegedly better at filtering any higher frequency stuff. My practice is just do this at the last cap in any network.

[quote author="zebra50"]
Something looks not quite right - 95V is lower than 115V but appears closer to the diode bridge. I know there's a 10Meg, but this doesn't appear to be part of a voltage divider. What's dropping the voltage? Either there is a resistor missing, or the two voltages are backwards, or...?[/quote]

I agree this looks fishy. Before I check the circuit again for mistakes, is it possible that the input impedance of my voltmeter may be a little low for measuring the voltage from this 10M resistor to ground? I'm not sure

[quote author="rodabod"]I agree this looks fishy. Before I check the circuit again for mistakes, is it possible that the input impedance of my voltmeter may be a little low for measuring the voltage from this 10M resistor to ground? I'm not sure[/quote]

Highly likely---measure the other end of the 10M if accessible. Most DVMs have about a 10M input loading on the higher voltage ranges.

I quickly redrew Roddy's sketches, although I forgot the output caps. I'll fix that later when we've sorted the rest out. Please check for errors.

It might be worth measuring the voltages at each of the 33uF caps.

:thumb:

Cheers, guys.

So, I've re-measured the voltages at the nodes above each cap, and from left to right, we have as follows:

230V, 196V (polarisation), 158V, 118V. The 0B2 is sitting at 109V.

That's a mighty high polarisation voltage.