DIY Factory

Kent's LA2 compressor Project
A project brought to you by Kent Stevenson
Updated 25 March 2005

The LA2 Homebrew Page

There are quite a few pictures so be patient if you have a slow connection (like me).

I started this project in sort of a roundabout way. I was mixing some tracks in good ol' Pro Tools and was getting frustrated with the vocals. They sounded pretty good in general. Tracked with a Neuman M147 tube mic through Focusrite ISA110 pre's. I don't usually compress until mixdown (that's changing) because I like the control of messing with it later. And they were 95% there. I was frustrated about that last 5%. Around that time I had been listening to some tracks a friend gave me of Chris Cornell (Soundgarden) singing. These were raw. No mastering. No effects. They sounded spectacular! I wanted THAT sound. My buddy, who worked at Bad Animals in Seattle, said they used an LA2A when he saw them track the stuff. I didn't have the jing for a real one and I had built a couple tube amps before. And there you have it... new project 2002.

I did some scouting around the web and came upon Rob's site for Classic Pro Audio. It gave me lots of great info like the manual, older schematic (with errors) and some parts sourcing.

Here is a parts list with some pricing. This has some of the sources I used but a few things (like the T4B) I had access to through my work.

The first thing I needed was a chassis. Middle Atlantic had a three space shallow rack box that was inexpensive and a perfect fit (part# EC3). The only drawback is that it's steel so it's not as easy as aluminium to work with. Nevertheless, I had some good Greenlee hole punches and a stepped drill bit that did the job well. Next I ordered the resistors and caps from Mouser Electronics with the resistors being all metal film for lower noise and the non-polarized caps were Orange Drops. The power transformer was Rob's recommended Allied Electronics part #227-0113 for around $20. Tubes and sockets were from Triode Electronics. The input and output transformers were Jensen's who has an excellent article on hookup modifications for the LA2. However, the big score for me was finding a couple of Urei T4B optical modules in the service department where I work. One was bad but the other was apparently working. I opened them up and took some pictures of the insides:

Above is an original Urei labelled T4B. You can see that it's a bit different. It has 3 LDR's (an odd one is wired in parrallel to one of the others) and also contains R8 which you can see toward the bottom of the photo. I would think that this is the difference between the T4A and T4B but the labels on both say T4B. There is a screen mesh which seperates the LDR's from the Photoluminecent material.

T4B's can still be purchased but for the real DIY'er there was a great article at the old Group DIY pages but alas it is gone now. Don't fret as there is always movement in the DIY world and help and wisdom is never far away. Set course for the LAB at www.groupdiy.com and look for the META threads on making all types of substitutes. The photoelectric material used can also be found in the meters of old MCI consoles and I found some and took some more pics. The label says "Grimes MFG. Co. Urbana, Ohio USA - 115V 60Hz". I haven't tracked down any more info on it although some of the "magic" is said to be in the use of this funny little part.

Here is a photo of the backside of the T4B that worked. Without clipping the cable tie I assume there are holes in the PCB for the light to shine through.

While I was waiting for parts I came up with a rough layout of the circuitry which you can see below.

_
Click on it for a bigger picture

It had a few errors and missed parts (which I fixed for this page) but it got me in the ballpark. You can see that I was going to do most of the wiring on turret board similar to a vintage guitar amp. In the end this worked out fairly well and matched up to the placement of the large parts.

I also did a some measurements for the parts placement of the back of the chassis

After some parts came in I made some simple paper cutouts of the same size and started laying out roughly where they would go in the chassis. I used pictures of an original from Universal Audio as a guide. I figured that they had spent plenty of time moving things around to get the leads respectably short and hum to a minimum so why try to second guess them!

Next I wired the turret board. A local electronics store was having a moving sale and had a bunch of pre-punched ones for $1 a piece so I bought a few for future use. These worked out great although it is possible to make these boards yourself. Wiring them before putting them in the chassis made things very easy. Below is a picture of the board wired up and the blank You'll notice the metal-film resistors for lower noise.

Here is a close-up. I sawed the board down the middle. Running along the bottom is the ground bus.

This is a picture of my first run at a power supply. My original idea was to have the heater supply run off DC. You can see the extra diodes for this at the top of the power supply. That didn't work too well (the voltage was a bit low) and I noticed no difference in noise when they were run AC so I took those out and left the supply the same as the schematic.

Here is a close up of the heater wires. A couple of things should be said about the these. First, it's pretty important to twist them tightly to reduce any stray noise from the high current AC. Second, I make sure they are run along the back corner of the chassis which acts as a bit of a shield

While I waited for more parts to come in I drilled and punched the chassis. My wife got me a used drill press last Christmas which was invaluable. The tube sockets and switch holes were punched with Greenlee punches. I love those things - perfectly round holes and clean as a whistle!. Some of the odd size holes were drilled with a step drill bit like the ones found at most hardware stores.

After I got the rest of the parts in I loaded the chassis and started wiring. Below you can see pictures of this in progress. With any tube gear the lead dress is very important. The general philosophy is not to run wires to the tubes parallel to each other. This explains the screwy turns in the picture. Keeping them short is always a good idea and I always use shielded wire to the tube grids with the shield grounded at one end only.

Then comes the payoff... well, almost. Time to fire it up. With any DIY tube gear I use a trick a friend showed me when powering up for the first time. Wire up a raco junction box with a couple outlets on it and put a switch in it. Connect the two outlets IN SERIES and wire the switch to take one of them out of the circuit. Plug a light such as a utility lamp into the switched outlet and your gear into the other. When you turn on your gear the lamp will glow a small amount for a moment on power up and then soften. If there are major problems with your DIY tube thingie and it starts sucking current in a big way then the light will glow bright and use it up giving you time to shut down before burning up a transformer or other major parts.

On it goes and so far so good. First I did a quick sound test and it passed audio. Next I measured the voltages at the plates. These were way off. After triple checking everything I turned to my friends at the Tech Talk section of Recording.org . Here I found some invaluable information and mostly have Byron to thank. Firstly, he gave me his voltage readings:

"The value for R25 is typically 33k.
typically voltages at pin 6 for me are in this range, 12ax7#1 - 100v, 12bh7 - 200v, 12ax7#2 90v, 6aq5 - 80v but up to 20% higher shouldn't be out of line. look also for incorrect plate resistors as these dictate the voltage drop to the tube. My la2's meters don't usually respond until about 11 oclock with typical signals. So, this is either normal or we are building them wrong. If you get both signal passing through and reduction and it sounds good, then you are almost home I'd say. Also, R8 is omitted on later la2's if you are using a t4b and adding r8, this may be a problem. r23 is 6.8k on later schematics. If it helps any, i just went crazy with my latest unit, false high voltages and no compression. Found I simply reversed the stereo interconnect causing a path to ground for the signal. Only wasted 4 hours!!! figuring it out."

And some later info:

"Also, check pin 3 at T4B socket. You should see up to 60 to 80 volts AC as signal is being compressed. The AC for the light inside the T4b comes from pin 5 (or is it6? sorry, I'm at work, away from the schematic) through a cap on its way to T4b pin 3. If you are getting voltage rising and dropping, from 0 to 60, maybe as high as 80, then all is well with the circuit and the problem is the T4b. If you really have access to failed T4b's then try and open them up get us some pictures. Chances are either a connection is bad or the light has failed. Easy to repair. As far as I know, R8 does not exist in a T4b. You say you pulled the case off? Off of the T4b? I could be wrong but I'm not opening my T4b to find out! C4 is a variable cap 50 to 350 pf I believe. Dave Jahnsen reccommends a fixed value of 100 to 200 pf as a good start. It controls the frequency response of the unit. lower value = more low end. So no reason you can't use the one you have and back it off a little or plug in a fixed 100 pf. I have a LA2 that I forgot to adjust the C4 before I closed it up and can't be bothered opening up. So, its at 50 pf and i think it sounds fine. I don't use C14."

When tuning the variable capacitor (which I left in) I found out later on a scope that it makes a BIG difference in frequency response in the area of several dB. Our testing gear will do a frequency sweep and display a plot which helped a lo. I would suggest getting an analyzer of somekind, software ones work well, and tune it that way. Mine ended up with a 1/2dB rise at the top end which was much better than the 6 or 7dB slope I started out with.

When it came to the voltage problems the other little tidbit I found was the fact that the schem in the Audio Cyclopedia is WRONG. R29 is supposed to be 4.7K like the later version. More Byron info:

"......here's a very interesting thought from lulu the spacecat regarding r29 and voltages inside the la2:
I asked about the 47k R29 LA2A resistor only 'cause when I built the first one the schematic I had (from the Audiocyclopedia) used it. Well, of course all the voltages were weird so I changed it to 4.7k and suddenly all was well. "

Once that was changed the voltages all behaved themselves. Hurray! Now I had to deal with the HUMMMMM. I connected pin 1 of the input and output directly to the chassis next to the jack. This is usually the safest place to ground those when it comes to noise. The case ground and center taps (B+ and filament) of the power transformer were connected directly one of it's mounting bolts. I made sure I scraped off the paint there so it made a good solid connection. I've seen other gear where they have just put a small round sticker over the hole before they have painted which leaves a nice little spot to connect the grounds when you remove it. Unfortunately I didn't have that option since my box was pre-painted (I still had to go back and paint it again anyway after all the scratches I managed to put on it). When I first wired the turret board I put a ground bus down the back side of the strip. Since my layout had the amplifier and gain reduction sections split to either side it was easy to seperate those ground connections. This left me with what is essentially a psuedo-star ground. All of the amplifier sections going to one ground bus then to the ground lug of C7-B and the gain reduction section grounds to another bus then straight to the transformer mounting bolt. That did the trick and the output had to be cranked to ear crunching levels to hear any hum at all.

I brought the box to work the next day and had our tech put it on his bench meter which can calculate THD, noise, meteriing levels tec. Low and behold I was at 86dB S/N which was better than the reissue model! There was no 60Hz or 120Hz above the noise floor so my grounding seemed to work. The only issue was with my VU. It was off slightly as I applied more and more gain reduction. The meter I had was labeled 7k impedence so I scrounged up one that was 10k from the junk parts bin at work. This along with 23K for R25 seemed to get me close enough to work with.

Here is the box finally assembled nice and neat.

After using it a few times and sending it out with a couple of pros in town
the results are in. and it sounds FABULOUS !

- Special thanks goes to Kev & Byron and the rest of the old Group DIY gang. Also Scott Cummings at American Pro Audio for his old junk parts and usefull knowledge. And Dan Kennedy for giving me an "atta boy!" when he got a look inside the box (that will make your head swell).

ERRATA

Kent Stevenson ©2005

* updated 11/15/02 - fixed layout & added grounding info

*updated 12/5/02 - added picture of finished box with lettering & logo

*updated 2/11/03 - fixed R3/C10/R36 connection to ground and the direction of CR1 on the layout (thanks Peter!)

*updated 4/30/03 - added price list spreadsheet.

brought to you by Kent and the DIY Team at :
www.diyfactory.com