etheory - a bunch of projects in progress

Hi there!

I've been extremely busy as of late doing a whole bunch of design bits and pieces.

Now, before someone chimes in and tells me not to re-invent the wheel, know before you do that I am well aware of this 8)
If I wanted something I knew already worked then I'd be happy to buy something (and have!).
However, I love design, circuit board layout, learning and making things, so this is just a whole bunch of fun and relaxation for me, and good circuit board layout is ART!

There are a few esoteric and interesting things in this bunch however, and I finally felt I'd made enough progress to share.

The projects I currently have running, at various stages, and in no particular order are:


API 225L/2500-ish clone:
A VCA comp.
Progress so far (only the audio path, but it does work, sounds awesome, and was verified as per my intended spec - next step is to work on the side-chain).
The DOA's are original clones of the 2510 and 2520 both in 2520 format:
http://www.evolutionarytheory.com/2012/04/custom-audio-compressor-%E2%80%93-part-3/



Siemens Sitral U274 inspired compressor/limiter (eComp01):
This one is based around a diode bridge-ish topology.
I'm mildly obsessed with the BC109C and 2N2219A, so they make an appearance too, though BC550C + BD139 would also be perfectly suitable (note the dual BC109C/BC550C footprints to cater to this).
The current circuit board design I've been working on and am hoping to prototype this weekend looks like this:


After the parts delivery for the prototype:


I went on to print out a size-test using a piece of paper and patiently pushing parts through.
This seems to verify the basic layout will work:



eComp02 FET compressor - at times I feel this one is a partially up-hill battle.
It appears nearly impossible to get less than 0.68% distortion with a FET design. Then again it's possible that this is the reason they sound so good also ;-)
Recently, having trialed nearly every topology I could think of and referencing against basically every other FET-based compressor in existence, I came up with something I was happy enough to prototype to the CNC stage and did a layout for it (the audio-portion anyway - the side-chain is still a thought experiment).
The backwards Ed Anderson 2622 seemed like a good idea at the time, but it's just not going to work unless your source impedance is 600 ohms or lower unfortunately (I really want a step-down here, so I'll use something else):



Siemens Sitral U295b EQ:
This one has me rather excited and is imminent after the U274 and FET comp above are tackled.
This is one of many working simulations of this circuit/topology.
It looks like it should work rather well with Edcor WSM's front and rear which would keep down cost.
I also want to modify it to use two sweep-able mids instead of one.




Most of these (the ones that cause me to pull out the least hair 8)) will be available at some point as kits for purchase, and some I'll just release gerbers and schematics for if people want, with and without parts and maybe even with some form of custom case. There is still a bit of work to be done, but the U274 is VERY close to being finalized, and my fingers are crossed that the work I do on it this Sunday will bring it up to a good level.

Anyway, more soon, keep on DIY'ing!

cheers,

etheory

Excellent work!! The best way to learn is to do!

Couple of quick questions:

1. I think I saw somewhere that you were cnc'ing your boards. What software are you using?

2. On your spice model, where did you get the nifty switch model? I've been having to use some creative layout for that, and it gets annoying having to do it for every switch!

Thanks!
Andy

I'll do my best to answer:

Indecline said:

1. I think I saw somewhere that you were cnc'ing your boards. What software are you using?

Click to expand...

I am a member of the Sydney Hackerspace called "Robots and Dinosaurs" - http://robodino.org/. The CNC machine was purchased and in maintenance as a part of their service.
They use the Mach3 software. In Eagle there is an export script you can run using "run pcb-gcode-setup" and it exports the relevant files for Mach3 to drive the machine.

Indecline said:

2. On your spice model, where did you get the nifty switch model? I've been having to use some creative layout for that, and it gets annoying having to do it for every switch!

Click to expand...

My full-time job is a programmer, so when I faced this issue I figured out how to write new models for LTSpice and created them myself.
I have shared these models here:
http://www.evolutionarytheory.com/downloads/LTSpice/

The ones that might be relevant to you are:
http://www.evolutionarytheory.com/downloads/LTSpice/Potentiometer.asy
http://www.evolutionarytheory.com/downloads/LTSpice/Potentiometer.sub

http://www.evolutionarytheory.com/downloads/LTSpice/Switch12Pos1Pole.asy
http://www.evolutionarytheory.com/downloads/LTSpice/Switch12Pos1Pole.sub

http://www.evolutionarytheory.com/downloads/LTSpice/Switch24Pos1Pole.asy
http://www.evolutionarytheory.com/downloads/LTSpice/Switch24Pos1Pole.sub

http://www.evolutionarytheory.com/downloads/LTSpice/Switch2Pos1Pole.asy
http://www.evolutionarytheory.com/downloads/LTSpice/Switch2Pos1Pole.sub

http://www.evolutionarytheory.com/downloads/LTSpice/Switch2Pos2Pole.asy
http://www.evolutionarytheory.com/downloads/LTSpice/Switch2Pos2Pole.sub

http://www.evolutionarytheory.com/downloads/LTSpice/Switch6Pos1Pole.asy
http://www.evolutionarytheory.com/downloads/LTSpice/Switch6Pos1Pole.sub

If you want to open up the models to have a look, the method is basically this:
To open a switch to A, reduce the resistance between the input and A to be very low, and the resistance to B to be very very very high.
Then to switch to B, flip the resistances.

The logic extends rather nicely to more and more poles etc.

Note within these models you can drive them via changing parameters using the notation:
POS={SOME_PARAM} (note that once you place one of these models in your schematic, you can right click on them to get to their parameters, then just find the line that says:
SpiceLine      POS=1 and change it to the above).

My convention is this:
Inside your schematic, add two SPICE directives:
.param SOME_PARAM 1
.step param SOME_PARAM LIST 1 2

The first defines a parameter to drive the switch position with.
The second defines the simulation to be run with switch position 1 and position 2.
This naturally extends to my other models of 12 and 24 position switches.

To get, say, a 24 position 4 pole switch, you can simple duplicate a 24 position 1 pole switch four times, and drive each instance from the same parameter, making them switch in lock-step which has the effect of increasing your pole count.

Indecline said:

Thanks!
Andy

Click to expand...

No worries! Hope I've helped.

Great job fellow Sydney sider!

I have recently taken on the fun/frustration of designing PCB's... still a long way to go though.

I see you're ordering parts from Farnell. Do you remember when they had free shipping? man, those were the days.

Good luck with it all, any plans to head to the white market?

Braeden

braeden said:

Great job fellow Sydney sider!

Click to expand...

Thanks! Now hopefully it all works, that's the next bit 8)

braeden said:

I have recently taken on the fun/frustration of designing PCB's... still a long way to go though.

Click to expand...

Once you get past that frustration, and make enough mistakes to start making hardly any, it's a truly awesome thing. It really is art. It takes time, there are a lot of iterations, everyone has a unique reaction or perspective on where those little tracks go, and, in general, it gets people thinking. It's truly a form of art this is under-appreciated I think. Please make sure you post your results!

braeden said:

I see you're ordering parts from Farnell. Do you remember when they had free shipping? man, those were the days.

Click to expand...

Yep, from element14/formely Farnell. Free shipping was one of the greatest things EVER (I too miss those days), (not to mention their unparalleled customer service, which is TRULY outstanding), aside from them just being in Australia and having so many things available from one website.

braeden said:

Good luck with it all, any plans to head to the white market?

Click to expand...

That's the plan most definitely. But I won't release anything until it's been very well tested, modified to be easier to put together and cheap as possible.
I'll absolutely keep posting stuff towards that goal.

Ideally I want to sell small kits, including parts, to a wide audience that might normally shy away due to project complexity or lack of documentation.
These will be heavily documented so nothing can go wrong, and simple enough that if it does, it should be easy to identify/rectify.

Well, the day was extremely successful.
After arriving at the Hackerspace to find no operational internet, I managed to figure out how to configure my phone with WiFi tethering and after 2hrs my Dropbox had synced and I was away.
By using a few new techniques, brand new drill bits and a bunch of exuberance, I was able to get 4 PCB's done today.

Here they are:

First up is an NTP M100 DOA I've been working on (during and after the CNC process - the new drill bits are SHARP and leave really really nice clean edges. I'd never CNC'ed anything that nice before):

Next up the e274 diode compressor prototype PCB:

And after finishing:


Next was the ePreamp02 - a variation on the ESP Project 66 Preamp, but with a few nifty additions and circuit modifications:

Then finally the eComp02 - a FET comp board - but this is just the audio path, the side-chain will be a future project:


The new technique of editing the pcb-gcode-setup.ulp script for Eagle so that it only exports a single path around the board, rather than the multiple passes it usually does took the etch pass time from about an hour and a half per board to around 22mins.

Brand new drills for the holes meant I could run the CNC at 140% it's usual speed to keep the drill pass down around the 12minute mark, and the mill at the end was around 5-8mins per board.
Probably the cleanest and fastest effort I've done yet.

I fully expected to be working until well into the night, but I'm home at 5pm, which gives me time to put together an e274 for testing ;-)

When I got home I did my usual steel wool + alcohol clean of the board, scrubbing the copper vigorously to remove all cruft and fingerprints, then sprayed with pcb protection lacquer, then left out side to dry. I'll pop out in a minute to grab them.

More soon.....

Oh man, glad to see it's FINALLY starting to all come together. Looks awesome. Can't wait to hear all this cool new stuff in operation.

A nicer photo this time from the 7D (others were taking using my crappy camera phone).

Post-lacquer-awesomeness:

Almost there with the e274 diode compressor, resistors next (I'd usually go for resistors first, but I didn't know if everything would fit nicely next to each other on the board so I first placed the parts with customized footprints and they all fitted nicely), but decided to call it a night and get it up and running tomorrow:

You lacquer them before soldering eh...i take it that the heat of the iron melts the lacquer and allows you to solder to the pads?

It's a special "solder-through" lacquer.
I've started using it since I've had issues with tarnishing the copper by touching it - even though I'm not a greasy individual!

Yep, it seems to just dissipate when heated, and doesn't negatively affect the soldering process (though it smells pretty weird....).

When I'm finished soldering I'll do a second coat of the usual protective finishing lacquer.

hey etheory, that's a lot of awesome projects here!!! Never heard of those Siemens things, diode bridge sidechain, could be the same taste than a Neve Comp? Anyway well done, and good luck on finishing them.

ben

evilcat said:

hey etheory, that's a lot of awesome projects here!!!

Click to expand...

Thanks!

evilcat said:

Never heard of those Siemens things

Click to expand...

This is another reason I'm working on this device, since it's not very well known, needs to be brought to a wider audience, and has a very interesting topology that people don't normally hear without spending a lot of money. I wanted to bring diode distortion to a much much lower price point, but still retain some level of performance and character.

evilcat said:

diode bridge sidechain, could be the same taste than a Neve Comp?

Click to expand...

I doubt my revision of this would come anywhere near a Neve of any sort. This is more of a compressor you'd use to beat your audio to a pulp.
The original was a limiter added to the Siemens Sitral U274 preamplifier (if you are really curious it's on Kubarth's site in great detail), and is known for being VERY dirty (in a good way).
I ripped the limiter section out and ditched expensive parts of the preamp like the transformers and the huge gain, in favour of cheaper parts and, for line signals, higher performance and less distortion than the original. I'm really quite curious what character this thing ends up taking on considering how much I've "improved" on the original design in a few key areas.

evilcat said:

Anyway well done, and good luck on finishing them.

Click to expand...

cheers.

OK, tantalizingly close now on the e274 - the build is finished but it's too late at night over here to wire up the peripherals, so that's for tomorrow....

In the meantime here is the chronological build log:

etheory said:

evilcat said:

diode bridge sidechain, could be the same taste than a Neve Comp?

Click to expand...

I doubt my revision of this would come anywhere near a Neve of any sort. This is more of a compressor you'd use to beat your audio to a pulp.
The original was a limiter added to the Siemens Sitral U274 preamplifier (if you are really curious it's on Kubarth's site in great detail), and is known for being VERY dirty (in a good way).
I ripped the limiter section out and ditched expensive parts of the preamp like the transformers and the huge gain, in favour of cheaper parts and, for line signals, higher performance and less distortion than the original. I'm really quite curious what character this thing ends up taking on considering how much I've "improved" on the original design in a few key areas.

Click to expand...

That's good news, we already have Neve stuff. Your improvements seems quite cool, I'm really curious with the results.
Best,

ben

Those PCB's came out super awesome! Very cool stuff. Thanks for posting.

OK, now the fun part, debugging....

Upon connecting everything, and turning it on, the U274 doesn't "work" 8) (unless you expect it to be a high-frequency oscillator, in which case it excels!)
But I am not discouraged. I will fix this thing! And it will be awesome!

Well, it works up until a particular point I've traced out, but then it goes a little bit mental.
I have a feeling it was rather folly to attempt to replace 180 and 120Hfe min transistors with 450-900 ones. I'm sure that's not helping the oscillation or stability situation.

There seems to be some high-frequency crud appearing at a particular point where there is some gain, so that's obviously a good first target.

On the bright side, the thing ain't overheating or producing totally confusing results (at leat not yet), so, at least I've managed to power it up with my eyebrows in-tact.

More soon once I debug this thing.

cheers

(all this is with respect to the schematic on Kubarth's site for the Siemens Sitral U274, so part numbers refer to that)

I have some exciting news.

I managed to get the amplifier section of the U274 amplifying. And it actually seems pretty good.
My method of using one side of an NE5532 to replace the input transformer actually works incredibly well.
In this context it replicates the functionality of the original transformer almost precisely, but with better linearity.

It seems that I must have calculated the the total emitter resistance I was using for Ts1 whilst drunk.
It was WAY off, which was messing with the gain and bias in bad ways.
I fixed that, then started playing with the trimmer, and then good things started happening.

I could see the parasitic oscillation on the oscilloscope still, but only during a brief period of the waveform. So, doing what any inquisitive child does, I put my fingers on everything, one at a time, and BOOM (eyebrows perfectly still in place), I got perfect amplification when touching either Tr2 or Tr3. The second I let go, I get high frequency fail reappearing.

The high frequency noise is definitely gain related, Ts2 and Ts3 are the main gain stages for the output.
Touching Ts1 makes things worse, so my assumption is that Ts1 is where the issue starts, then Ts2 and Ts3 have their bases shunted to ground by my body as a capacitor, effectively filtering this noise out as a low pass filter.

I FINALLY managed to make the simulator react exactly the same as the circuit, including parasitic oscillation, and a 47pF cap between B/C in Ts1 stops that right in it's tracks, it would appear.

Regardless, I now have a clear path to follow and an amp that is finally amplifying (even if it requires to be touched to work properly, it has feelings too you know).

Great news.

More over the next few days....

When you say things like Tr2, Tr3, Ts2 or "method of using one side of an NE5532 to replace the input transformer"... I can't say I have a clue of what you are talking about. Did you post a schematic at some point?

The schematic for the original U274 is available from http://audio.kubarth.com/rundfunk/index.cgi.

I'll post my redraw of it when I get home from work.
That'll show the opamp replacing the transformer thing I'm talking about.

thanks