Microphone Preamp design

[Michael Koster]

I am struggling to understand this. 33dB is about 45 times so the plate feedback resistance should be in the region of 45 x 680 = about 31K. Since in triode mode, the mu of a D3A is about 80 the feedback resistor 'looks like' 31k/mu from grid to ground or about 390 ohms. This plus the 680 gives an effective input impedance of 1070 ohms, a lot less than the 1600 you calculate. As far as I can see the '65 ohms for the triode' (whatever that means) and the rbb of the bias transistor don't come into it. Or have I missed something??

Cheers

Ian

Hi Ian,

Here's what I'm thinking and thanks for checking my work ;-)

I'm assuming mu=77 from the datasheet for the moment.

So the part of Rf that's seen at the input added to Rin is correctly Rf/mu

therefore the gain expression is: Av=Rf/(Rin+(Rf/mu)) (right?)

which results in 72K ohms Rf and 680 ohms Rin for gain of 33 dB.

Zin = Rin + (Rf/mu) as you have, but with 72K you get 1615 ohms.

Does it also follow that the noise resistances are all in series? I assume those include the reflected source resistance, the reflected primary DCR, the secondary DCR, the input resistor, the ENR of the D3a in triode mode (65 ohms) and the ENR of the Vbb of the darlington transistor I use for cathode bias.

I'm still in the process of validating all the settings and specs after putting my breadboard into the package with the selectors and all, so there could be some surprises still...

Cheers,

Michael

 

[Brouilly]

Helloooo,

So I am finalising my proposal at the moment and I had a thought...

I am going to be in a position over the next 3 years where I can thoroughly involve myself in this preamp design in the lab, and I was wondering if anyone here has any unfinished business so to speak.

I'm sure this is when you tell me that everything has been done, as I'm sure it has. But is there anything that you would be interested in knowing? Something that you havn't had time to test?

This M.Phil for me is not just about the preamp but also about me really understanding valve circuitry, and any extra testing I can do will help, even if it's a bit off topic. And if everything has been done, can anyone like Michael Koster and Ian etc etc that has done this before maybe suggest an area of the design/research that they found interesting when they were on the same path as me?

Thanks for any input

Charlie

 

[ruffrecords]

One of the real problems with tube mic pre design is closing the NFB loop at dc whilst being able to vary the closed loop gain over a reasonably wide range.

Typical tube mic pres using pairs of triodes with global NFB around a pair always have a blocking capacitor in the feedback arm because the output dc level is usually far higher than the input (or feedback input) dc level. The problem is this means the closed loop gain rises at very low frequencies and there is often an (unstable) peak. What is more, varying the closed loop gain by changing the feedback only makes this peak worse as gain is reduced.

Semiconductor designs have no problem with this as it is easy to arrange the input and output to be at similar dc voltages and the loop can be closed at dc so there is no rising response at very low frequencies.

I have seen one (Pultec) tube mic pre design that does close the loop at dc but the the range over which its closed loop gain can be varied is limited.

I would therefore be very interested in topologies that allow both closing the loop at dc AND the ability to vary closed loop gain over a wide range (say from 0 to 40dB).

Cheers

Ian

 

[Wavebourn]

I did DC coupled mic preamp with tube input, FET out. +250 and -72V were used. People said it was not kosher, but sounds like a charm.

 

[ruffrecords]

I did DC coupled mic preamp with tube input, FET out. +250 and -72V were used. People said it was not kosher, but sounds like a charm.

Well, of course, as soon as you allow semiconductors to be involved a whole new range of possibilities opens up. I was thinking of tube only topologies.

Cheers

Ian

 

[Wavebourn]

Well, of course, as soon as you allow semiconductors to be involved a whole new range of possibilities opens up. I was thinking of tube only topologies.

I invented a new term for it, it is called now "SS Assisted Tube Amplifier". It is contrary to "Tube amp made cheaper because of SS devices to cut costs".

 

[Brouilly]

Hello all,

After much procrastination I have started my study. The first thing I am doing is looking at a very cheap mic valve preamp that I wont name. It cost around £50 new, and has two pretty little ECC83s on top.

Ive taken it apart and traced most of the pcb and it is very strange indeed. The fact that only one half of each valve is being used tells me that it is a gimmick, but there is some circuitry involved. It appears that both valves (that are used) have their grids connect to ground via an RC stage. I still cant upload photos to this site, but it is essentially.

Ground - 10K+2.2uF//10K - Grid.

Ive seen grounded grid topologies, but this seems very odd. Especially as the anode of one of the valves isnt connected to anything.

So is this all just for looks? And a waste of components, infact there is a pcb track all the way to one anode pin, but that half of the valve isnt being used? All seems quite silly to me.....

... I guess it glows in the dark though....

Charlie

 

[DF96]

For many people ECC83 is now a fashion accessory. My guess is that for £50 all you get is a fairly standard op-amp (and not necessarily even a low noise one) in a box, with a fancy (and not necessarily safe) power supply. I assume this is a Chinese-made item, although not necessarily Chinese-branded.

I assume you intend to design and build a genuine valve mike preamp, not a spoof version like many of those currently available?

 

[Brouilly]

I assume you intend to design and build a genuine valve mike preamp, not a spoof version like many of those currently available?

Absolutely. I am only looking at this to see how they have used the 2 ECC83s for £50. My tutor chucked this preamp at me a few days ago so I thought Id take it apart. I think the B+ is supposed to be somewhere around 50v (toriodal transformer is 15-0-15), that im sure comes in handy with the phantom power. But the one I got (fresh out of the box) only had a B+ of about 4v..... and the phantom power never got over 2v.

My only explination for it is, that if someone opened up the box and saw a few resistors and caps near the valves, they might believe that they are being used.....

Better not use this as a benchmark preamp aye?

Cheers

 

[Brouilly]

I recently visited a fellow DIYer to buy some of his workshop off him, and the majority of the valves he is selling are DHTs and an interesting conversation started about DHTs in recording studio equipment.

Were DHTs ever used in mic amps? I have been told (not yet built/heard an amp using DHTs) that they are sonically superior, so surely they must of been tried?

I understand that there are a number of problems using them - space they take up, little gain, poor microphony. But I am sure with clever techneques these problems could be solved.

I recently read in the RTP5 thread that Shoog was looking at using a 3A5 at the input of the phono stage, so these ideas are bouncing round. Has anyone had any experience with this? Or is it really not worth it?

Cheers

Charlie

 

[piano3]

Charlie, using a DHT for a line stage can be tricky enough. ECC83 should be fine for an input stage but you will definitely need a low output impedance follower to interface with your recording device. My most recent micpre uses a high gain UHF single triode with a trioded E180F cathode follower output. If you don't want to use a follower you might try 2 stages of low voltage, high current and high gm and low microphony single triode 6s31b(mu about 18,look on ebay) which if you choose your operating point right should give an output impedance of about 1k. Or else, 2 stages with higher total gain and a stepdown transformer.

 

[Wavebourn]

My most recent micpre uses a high gain UHF single triode

6С17К-В?

 

[piano3]

Hi Anatoliy,

Yes indeed; the trickiest part for me was designing the heater supply. I eventually used(after trying various things) a simple CCS supplied zener with Darlington emitter follower output but then supplied the filaments via a cascode BD136 CCS. This seems now quiet enough.

 

[Wavebourn]

The trickiest part to me was powering of 8 filaments, when cathodes of mic inputs are connected to filaments. I found too that separate CCSs is one good way to go.

Also, I adjust bias by grid stoppers' value. Tiny grids are so close to cathodes, they work like power stations. 😉

Nice tubes. But quite different from the rest of triodes.

 

[piano3]

I have the 6s17k actively loaded as well passing just a few mA. I used just 4k7 at the grid because that is what I usually load the mics with. Have you found any benefit from going much higher than this?

 

[Wavebourn]

I have the 6s17k actively loaded as well passing just a few mA. I used just 4k7 at the grid because that is what I usually load the mics with. Have you found any benefit from going much higher than this?

It depends on tubes. I use input transformers, and 1-10k in series with secondary to grid. Yes, more around 4k7.

They have wide spread of parameters, including input currents: 8 micron wires in grids, 18 micron between wires. They were assembled in vacuum by robots, I suppose it is quite tricky to make them with desired precision.