Ok, so I got some more testing and things done. The unit powered up and the voltages seemed as expected! I should finish wiring and give the first one a test tonight and then it should be fairly easy to continue with the rest.
However.. I completely underestimated the amount of time it would take me to wire up everything. I really should have spent more time in redesigning the PCB to accommodate PCB mounted jacks and such, but I imagined it would only take a little while to wire this stuff up.. it's taking about 4 hours a piece to stuff, solder and wire up, in addition to the hour to drill/build the chassis.
As such, I'm going to have to raise the price to 300$ a unit, which still doesn't really pay for all my labor, but should be a good compromise.
Got the first one finished! Put it in the rack in place of my old 9K preamp to see how it would sound in the same place..
I have been using mine for overheads and snare for a while due to the crispness.
Here's a audio snippet. It's a full drumset for about 20 seconds, then cuts to the snare and overheads. All I did was mute the rest of the drums for the last few seconds.
The snare still has compression from my 4K gate/channel compressor as well as a little SSL bus compression and some 33609 parallel compression. It has tons of EQ which I didn't pull off either. Overheads have no compression and HPF around 400, and some cut around 3K and some light rolloff above 12K.
My original plan was to try the taper bending to turn a linear pot into an anti-log pot. Antilog is where the value of the pot increases very sharply for part of the turn and then slows down during the rest of the turn. The purpose is to counteract the logarithmic nature of volume when gain is increased in order to make it *sound* like it's increasing in a steady linear fashion.
So i started with 2K pots, which are the closest to the 2.2K pots originally used, but the taper bending resistor greatly influenced the resulting resistance value. Normally this isn't a big deal in a *divider* topology since in that case you're only interested in the current's value, but the way this is set up is a rheostat, so the resistance is also important and it comes in too low.
So I did find a "fix" for this and it's going to be replacing the 2K pots with 10K pots and using the taper resistor ratio to end up with the final value of 2.2K while still giving an Antilog resistance curve.
I'll swap out the first unit's pots and test probably friday night. Tonight is date night and the missus won't let me muck around in the laboratory tonight.
Otherwise I've enlisted the help of my cousin to help stuff PCBs, so I'll be working on wiring while he stuffs them to cut down on the time. He owes me a few favors but I'm going to end up paying him so I can get this project moving.
So I have good news, and bad news and good news and bad news and maybe some more good news..
Good news is that I have all the boards, stuffed and ready for installation.
Bad news is that I made a rookie mistake I don't think the anti log pot trick will work here, so we're stuck with linear pots for gain.
Good news is that they don't react all that different from my true SSL antilog pots in the unit, so maybe it's not a big deal at all. In usage, the SSL preamp has so much gain, even at lower settings, that I rarely turn it up more than halfway for just about any source anyway. You guys tell me what you think about this.
Bad news is that during my experiments I found the DC biasing to be very wrong for the MAT02 transistors. At full gain, the DC goes all wonky and the sound pops and cuts out and the DC level goes haywire.. During investigation I found that the two sides of the MAT02 were giving me drastically different measurements.. So I quadruple checked the build and found no stuffing errors. This led me to test a few of the new MAT02 out of the batch I bought, and the HFE is not matched, nor is the currents under usage..
So that led me to pull one of my known-good MAT02 from my SSL preamp I built 10 years ago, and measure it. Perfect matching, and when I put it in the new PCB, it works perfectly.
I think I got sold counterfeit parts! I bought these from a reseller, which I'm going to complain to, but ultimately, I think I'm going to have to eat the cost on these and go a different route. I can't afford to eat the cost on brand new MAT12s from Analog as they've marked them up to almost 30$ a piece.
However, the good news is that they have a part called the SSM2212 which is essentially the MAT12/MAT02 in SOIC format. The footprint on a narrow SOIC almost aligns with the pins of the TO-39 footprint to the point where I should be able to solder them on without much fuss. They're also substantially cheaper than the MAT12 in TO-39. I'll buy a few and see how they work.
This thing has been snakebit from the start.
I'm starting to rethink doing 5 more of these at all. If I do, I'd need to redo the PCB to cut down the amount of wiring, and to fix a handful of fitment issues.
So I cut one of these "MAT02" open, and aside from the fact that it has some kind of metal cap over the real top of the transistor, it has two single dies instead of one die with two transistors.
This furthers my belief I was sold counterfeit parts as you can't match two separate transistor dies like this.
I've sent a complaint to the reseller, but we'll see what happens. I'm sure they'll claim ignorance and say they bought from another, etc.. This is how it goes with EOL parts and resellers.
That’s gotta be really frustrating, sorry to hear it.
Absolutely frustrating, but it's a fact of life in the electronics industry now. You can buy a generic dual NPN for pennies, have a sweatshop worker in china or india stamp out little covers and print whatever you want on the top and sell them for 1000x what they're truly worth.. Usually by the time the end user discovers the counterfeit, it's changed hands through resellers a few times and nobody will take responsibility.
However, this reseller has engaged and wants to talk about returning the parts, which is good. Usually they flat deny anything and pass the buck to their supplier, etc.
I ordered some SSM2212 parts from Mouser this time to see how they work. In the datasheets, they're almost identical to the MAT02/MAT12 specs, just in SOIC form.
So the reseller was seemingly open to me returning the parts, but has since stopped replying after they realized I'm outside their return policy window, so the jury is still out whether I have to eat a few hundred dollars in counterfeit parts. I figure they'll use the policy to stiff me rather than do the right thing, but I'm open to being wrong on this..
I think I might do another run of these, but change the PCB to get rid of most of the wiring and incorporate the SSM2212 parts instead. Functionally it'll be the same as the first 5 hand-wired versions though.
I've found a source for dual 5k revlog pots, so I ordered a few to try in place of the 2.5k revlog pots.
Normally this would be an issue if the pots were used in a voltage divider arrangement, but this is being used in a rheostat arrangement, so the total impedance might not matter.
The other thing that makes this possibly work is that the least resistance is the most gain, so total maximum gain stays the same and we might get less gain at the bottom end which might not be a bad thing, especially without a pad.