Active Ribbon Preamp/Impedance Converter Idea - opinions?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Rossi

Well-known member
Joined
Dec 11, 2004
Messages
1,532
Location
Germany
As some of you may remember, I conducted some experiments in building and rewinding ribbon transformers:

http://www.groupdiy.com/index.php?topic=18764

http://www.groupdiy.com/index.php?topic=19046

My main goal was to build an "overwound" transformer with a very high step-up ratio of at least 1:80 and use an active impedance converter to get back to a usual mic output impedance of 50-200 ohms. My rewound transformers work quite well. Now comes the electronic part. For my first experiments I used the Schoeps/Dorsey circuit. Actually, I only used the hind part, feeding the transformer signal into the coupling caps:


simplecon.gif


This works okay. The resulting active ribbon is about 12 dB louder than the same model using the stock transformer. Noise performance is about the same. As some commercial active ribbon designs are lower noise, I figured, there's room for improvement. So here's what I've come up with so far:

aribbonpre.gif


The front end is a FET input à la KM84 followed by a BJT phase splitter. The FET part is configured for an extra gain of about 12 dB on top of the transformer voltage gain. The resulting output is pretty high, maybe a little too high, about as high as the hottest condenser mics. According to my preliminary tests, noise performance (FET bias hand adjusted) is a little better than the Schoeps/Dorsey-derived ultra-simple design.

So everything looks fine. Nonetheless I'd be extremely grateful for some input from you guys. This design is a little unconventional, I guess, and I'm not really an electronics engineer. So there's a good chance I missed certain things. I'm sure there's stuff I should revise. Maybe my design is crap altogether. :? You never know...

Opinions anyone?
 
I've thought about doing this before. I have quite a few ribbon to Hi-Z transformers.

I'm not sure about the performance of these higher ratio transformers though as it can have a detrimental effect on the audio. I should test some of the ones which I have.
 
What kind of Hi-Z transformers are those? Stock transformers from old ribbons? There seem to have been quite a few Hi-Z ribbons back in the day.
My rewound transformers sound pretty good, no worse than the stock low-Z transformers.
 
Nice all this :thumb:

Not sure though about the JFET w.r.t. noise. What's the estimated source impedance presented to the JFET ?

|Ribbon-Z|*80*80 perhaps still better tackled with a BJT-input.

Regards,

Peter
 
You may have a point there. The ribbon has a resistance of about .4 ohms, my newest transformer is about 1:100, I'd say (I don't know the exact figure). So the impendance presented to the input is probably around 4.5K (including the transformer DC resistence.) Could be a little higher or lower, so maybe 3-6 K. I'm actually thinking of winding another transformer with an even higher ratio.

On the other hand, I know that commercial designs do use FETs. Also, one of the preamps I like best on ribbons is a FET-based design: UA Solo 110.
 
[quote author="Rossi"]What kind of Hi-Z transformers are those? Stock transformers from old ribbons?[/quote]

Yes. I've never tried them before.

You could also try a standard trafo followed by a buffer and then say a 1:4 after to provide balancing and gain. Perhaps a little expensive though.
 
With the stock transformer, input noise is an issue, as the source impedance is very low and the signal is weak. Which means you have to get very low Rbb transistors. PRR once presented a pretty nice ribbon booster circuit. I built one with exotic ultra-low Rbb transistors which performs well. But with less exotic (still low Rbb) transistors, noise performance is no more than okay. I'm pretty certain that a high step-up transformer is basically the way to go. The remaining question for me is the electronic part.
 
As some commercial active ribbon designs are lower noise, I figured, there's room for improvement.
How did you came to the conclusion (i.e. how did you measure) that commercial mics are lower noise?

If your 4k5 output impedance figure is right a 2SK170 would provide very low noise indeed. Alternatively any decent BJT biased at a proper collector current will do almost as good at lower distortion levels. I think I'd adjust the input stage for 0 dB gain (source/emitter follower)--provides much lower distortion and more reasonable output level. If you use the Toshiba part indicated you might add a bootstrapped cascode to lower distortion even more.

The 22 nF RFI protection caps at the output look like a pretty hefty load--equivalent to about 100 m cable... 1 nF might do the trick.

Samuel
 
Thanks Samuel! I recently reviewed an active ribbon which was indeed lower noise than a regular ribbon at a very low noise preamp.

I actually tried the 2SK170BL first. I have a spice model for the Linear Systems counterpart, but for some reason it doesn't load in my program anymore (it did initially). So I used the 2N3819 model instead. When I built the circuit, I couldn't get the 2SK170BL to bias correctly. Source voltage was below .5V, no matter how I tweaked R3 and R5 (I put in trim pots for both). I then put in a 2N3819; the ones I bought recently are very low noise. I adjusted both resistors for about 1.3V source voltage and 10V at the drain.

Yeah, maybe the 22n at the output are a little high, although, according to spice, the frequency response remains flat to several hundred kHz. For the time being, I haven't really put in any RF caps at all.

I do like to have some gain. 12 dB seems a bit much, but 6 dB would be pretty cool. I'm building this not just for my own use. I'm preparing an article on ribbon mic modifications. There's a lot of people out htere who don't own a really low noise preamp, so decent output level is a plus. I'd like to build a ribbon that behaves somewhat like a condenser mic. Immunity against P48, low noise and decent sensitivity.
 
I am afraid you need one more transformer for output if you want your mic to be state of the art.
 
A second transformer is out of the question for the project at hand. I'd like to have easily available parts. Soundwise, too, I would actually prefer a transformerless output in this case. I like transformers, but I'm not sure I'd want half a dozen of them in the signal path.
 
[quote author="Rossi"]I actually tried the 2SK170BL first. I have a spice model for the Linear Systems counterpart, but for some reason it doesn't load in my program anymore (it did initially). So I used the 2N3819 model instead. When I built the circuit, I couldn't get the 2SK170BL to bias correctly. Source voltage was below .5V, no matter how I tweaked R3 and R5 (I put in trim pots for both). I then put in a 2N3819; the ones I bought recently are very low noise. I adjusted both resistors for about 1.3V source voltage and 10V at the drain.

[/quote]

Pinchoff voltage on the SK170 is very low, usually. It is so much quieter than a 3819 that in many cases a special effort to bias it properly is warranted, IMO. But then if you are dealing with thermal noise in 4.5kohm (about 8.6nV/sq rt Hz at 25C) the benefits may only be at very low frequencies where the FET excess noise rises a lot.
 
[quote author="Rossi"]A second transformer is out of the question for the project at hand. I'd like to have easily available parts. Soundwise, too, I would actually prefer a transformerless output in this case. I like transformers, but I'm not sure I'd want half a dozen of them in the signal path.[/quote]

In such case I'd recommend to power your mic from internal battery. You may use phantom power to switch on your battery power.
 
@ Wavebourn: nope, battery operation would be impractical.

@ Brad: I'm glad you tuned in! Does the circuit look okay to you? The topology is a little unusual, I suppose, and I feel a little insecure as I have no EE education.

What bias voltages would you recommend for the 2SK170BL? Low frequency is not much of an issue, from what I can tell. Noise in the upper frequency ranges is a little more bothersome, although the overall performance is not bad, I'd say. I just had a friend in my home studio for our weekly session, and I could hardly get him away from my unfinished mic with its dangling veroboard. He was quite fond of the sound of his own voice :wink:
 
Thanks, Brad! :thumb:

As I said, this circuit is meant for an article on ribbon mic modifications (in a musicians mag).

So my main goals were:
- sensitivity & handling similar to a condenser mic (phantom powered).
- low noise
- simplicity (many readers may never have used a soldering iron)
- easily available components
- low cost
- stop global warming & establish world peace. :grin:
 
[quote author="Rossi"]
- stop global warming & establish world peace. :grin:[/quote]

I guess it needs a switching regulator then! :razz:


But regulators aside---one drawback of the circuit as shown is the inequality of output impedances: the emitter Z is quite low and the collector Z is roughly the paralleled resistors it's hooked up to, including the phantom 6.8k. So there will be sensitivity to disturbances induced into the line as it stands (that the receiver won't be able to reject unless it is configured in a compensatory way). If the cable run is short and the environment fairly benign, usually no problem, particularly as you have a hot signal.

Some other details: since you want the FET gate at ground, you may as well tie it directly to the secondary of the input transformer and eliminate C4 and R2. You might want to leave a spot for a small value R in series with the gate, and/or a ferrite bead, in case your FET wants to oscillate.

I don't see what R10 does for you other than rob the already-starved zener D1 of current, so it can go away. Also it seems the junction of R16-R17 could go directly to C7.

For maximum voltage swing on Q5 collector and emitter, the optimal bias will be such that the total voltage between C7 and the top of the zener D1 is split, w.r.t. the collector and emitter of Q5, about 1/4 - 1/2 - 1/4, and to achieve this the ratio R7/R6 should be closer to 3 (the precise value will take Vbe of Q5 into account).

There are several ways to go about equalizing the output Z's on each line out if that is worth pursuing. They do add parts though.

One of the simplest would just add an R in series with the emitter to C5---that's not a bad idea anyway as the output cap C13 tied ~directly to Q5's emitter is a recipe for potential high frequency oscillation.

A more complex but more symmetrical approach would have emitter followers for each output line.

As far as using the 2SK170, it tends to have impact in a number of areas. You need more current to operate it optimally, and then you have to make R3 quite a bit smaller. Then the impedance at that node being so much lower requires that C2 be a lot larger for bass extension. You also really want that cap to be larger than just the required amount for low frequency rolloff so that it doesn't cause distortion. But then the signals are still pretty small, so it doesn't need to be gigantic I suppose.

With a sim of an adjusted circuit using the 170, I was seeing a lot of gain, about (EDIT:) 35 at the drain and hence 70 when you take the additional factor of two from the "phase splitter" into account. The configuration is R3 = 120 ohms, C2 ≥ 220uF, R5 = 2k, R10 open, C3 = 3.3uF, R8 = R9 = 1k, R6 = 13k, R7 = 33k, R4 = 0 (R16-R17 tied directly to C7). No promise this is optimal but it looked reasonable in terms of distortion and frequency response. I left your output caps off---if used they should be a good deal smaller as Samuel suggests.

Overall, I suspect that you are going to have too much gain for some applications, based on what PRR told me about some signals he's seen. But perhaps it will be o.k. for moderate sound pressure levels.
 
[quote author="Rossi"]As I said, this circuit is meant for an article on ribbon mic modifications (in a musicians mag).[/quote]
I realize I should quickly place a Thomann-order before the to-be-modded types will all be gone :wink: :green:
 
Thanks a lot, Brad. :guinness: :guinness: :guinness:

The circuit looks much better now:
aribrev2.gif


According to spice, 2.2u ought to be enough for C3; -3 dB point is at about 11 Hz. I managed to reactivate the Linear Systems LSK170 model, althouh I don't know how accurate it really is (my 170s are Toshibas). I haven't built the new version yet, but according to simulation I get lots of gain. 27 dB on top of the transformer gain and the phase splitter. So this active ribbon would be more than 45 dB louder than a normal passive one. Maybe a little too much :green:

Acutally, I'm glad you mentioned the phase divider gain. I had totally forgotten about that. :oops: 5 or 6 dB gain would be enough, I think. So a source follower configuration, as Samuel suggested, would probably make more sense than my original plan. Would the 2SK170 still be superior to the 2N3819 as a source follower?

@ Peter: Don't worry, the big T recently got a new shipment. The new ones have "normal" ribbon transformers (about 1:30). For a while they sold a version with an "overwound" transformer (about 1:80) that could be used for an active ribbon. No rewinding necessary. But I like to think that my rewound transformer is a little better. :wink:
 
According to spice, 2.2 uF ought to be enough for C3; -3 dB point is at about 11 Hz.
Did you verify this with a suitable output load (e.g. 1k)? Not sure what current gain the phase splitter got but the load might get trough and increase this frequency.

Would the 2SK170 still be superior to the 2N3819 as a source follower?
Noise wise for sure. Note however that you actually would like to have gain as early as possible. This would ask for a different output stage topology though. Personally I'd go back to the Dorsey circuit I think (miccircuit.gif). Or do you see a specific advantage of your circuit?

I recently reviewed an active ribbon which was indeed lower noise than a regular ribbon at a very low noise preamp.
And how did you measure the microphone noise? Looks like a relatively difficult thing to do.

Yeah, maybe the 22 nF at the output are a little high, although, according to spice, the frequency response remains flat to several hundred kHz.
Frequency response is not the problem, but the capacitors present a load to the output stage, causing frequency-dependent distortion.

Samuel
 
Back
Top