[2006-01-17]

So I have been wanting to go beyond playing music, and try to create music. Unfortunately I don't play any instruments. I have attempted to play the guitar and the drums, but am terrible at both. So I thought maybe I could create something using some of the very cool software that is out there. And my wife Nyree is going to sing. A band is born :)

There is no better excuse to build equipment than a band, so I am excited about that for sure. I got a microphone, but need to amplify it to get a line level signal. So why not build a microphone preamp? How hard can it be?

I started by researching all of the old tubes preamp schematics I could find. I settled on the Altec stuff, partially because of the Bottlehead modified Altec mic mixer on the Botthlehead site, and partially because the Alecs use triodes instead of pentodes, and I would prefer to use tubes I have than have to learn about too much new stuff at once. The Altec mixers typically have 2 or 3 gain stages, usually 12ax7, and then a driver stage of some sort, usually a cathode follower. Well, I have never been a fan of cathode followers, and since these things use output transfomrers, any output stage that can drive a 15k load should be just fine as a line driver.

Schematics: Altec 1567A tube mic mixer. Altec 1566A mic preamp.

I decided to base my first try on the 1567A. I also wanted a headphone amp, so I basically built the first 2 stages of the 1567A followed by a pair of 6sn7s into UTV A-25 output transofmrers, which are similar to the Altec 15065 used in the original schematic. One of the UTC transformers drives a line output, the other drives a headphone output. I decided to omit the tone controls, omit the feedback loop, omit the first gain stage, and use a 6sl7 instead of a 12ax7. So yeah, by this time it was a completely different design. I also wanted to use a bunch of nice UTC transformers I had scrounged. I used a UTC A-11 as the input. This has a 50k secondary and either a 50 ohm or 200 ohm primary. Very similar to the Altec/Peerless 4722. The output is a UTC A-24, which has a 15k primary and many secondary configurations from 500 ohms down to 50 ohms. Very similar to the Altec 15095. I was planned to use the 50 ohm output for the headphone channel.

The chassis was a lot of fun to build. I like working with rack cabinets. I didn't have a punch or hole saw big enough to drill the hole for the panel meter, so I used this arbor cutting blade thing. It was noisy as hell and took a long time, but worked great.

It took me a long time to figure out how to implement the panel meter. First of all, I knew very little about dB, and the meter I had was not the standard configuration where 0dB = 1mW. It was an older style of meter that is configured to read 0dB at 6mW, and has a 5k internal resistance. I ended up making it work by hooking it up in series with a 10k resistor (for a total of 15k) across the primary of the output transformer. I calibrated it so that +4dB output (1.23V) reads 0dB. This is the standard for balanced interconnects. Incidentally this style of meter is considered obsolete for audio monitoring, but I had done too much work not to use it! I even broke the glass in it and had to find a meter with intact glass to transplant the guts of this one into.

I ditched the headphone channel almost as soon as I started building it. It was clear that this was going to be complicated enough without worrying about the headphone stuff. Plus, when recording vocals, I will want to mix the music, plus the singer, and this was only set up to monitor the singer, and so was somewhat useless. A seperate headphone amp would make more sense. Maybe someday. I finally got the thing built up and was really excited to try it out. Well, it worked great - if by "great" we mean the ability to output almost 20 volts RMS of noise. What the hell?! I guess you need to do a lot more to keep these things quiet when you are building such a high gain amp designed to amplify less than a millivolt of input to high level +4 dB studio output.

To see what sort of tricks you need to employ to make one of these things quiet I consulted the photos of the Bottlehead modified Altec mic mixer (original site here). Looks like DC filaments for sure. Also lots of current sources, including a shunt regulated power supply. Very deluxe. Doc Bottlehead uses a 6s4 driver stage instead of a cathode follower, so I'm not out of line there. One trick I really like is the snubber capacitors on the filament pins. This keeps high frequency crap picked up by the filament wiring out of the tube. I thought that shielded filament wiring would be necessary, but it isn't used here. Doc also used grid stoppers - resistors attached to the grid pin of the tube to form an RC circuit with the internal capacitance of the tube to reject high frequency noise and prevent oscillation. Nice.

Doc Bottlehead's beautiful Altec mic mixer modification. See the whole story here.

After implementing Doc's tricks I was again excited to try this baby out. Lots of hours in it so far, and I was ready to hear something. Powered it up and got 20V of noise again. Ok, now I am feeling like a total hack. What the hell is going on here? Then I remembered my 6c45pi spud amp and how badly it oscillated. The fix for that amp was to add resistance across the secondary of the input transformers. I think this is a result of having a large AC impedance between the grid of the first tube and ground, and allowing it to soak up any electromagnetic interference that comes along. I don't totally understand it, but I added a 100k resistor across the secondary of the input transofmer and suddely there was silence. Well, at least the noise was just hiss and a little bit of hum instead of the meter-pegging crap I had before. Swapping tubes for quiet RCA JAN 6sl7 and 6sn7 tubes helped bring the hiss down to a tolerable level. It was encouraging that a new Sovtek 6sn7 was only slightly noisier than the old milspec tube that was the quietest I could find.

Now I had a preamp with hum and hiss that was barely acceptable, and WAy too much gain. I was using 2 6sl7 gain stages, and a parallel 6sn7 driver stage. It you work backwards from the output at 1.23V, the sensitivity was around 0.0035mV. Um - gulp. That is a lot of gain. Most mics output at least 1mV. Even with a lot of that gain coming from transformers, no wonder it's so noisy.

To try to reduce the gain and noise a bit, I plugged a 6sn7 into the first stage. That helped a lot, but it is still a lot of gain. At least it is usable as-is, and I have gotten some good recordings into my laptop. And I have learned a hell of a lot.