Feeler: Quad 8 444 EQ with M/S extension in 51x format.

[Upacesky]

2 for me too

 

[dmnieto]

I am still going ahead with the routing of the board ( I had not had a lot of time lately). I am trying to fit everything on a single PCB and it seems that it might be possible. The only exception is the booster board for the 16V to 28V conversion that will be in a daughterboard. It is based in the same IC that Matador uses in his Ai booster, but with two of them and a PNP level shifter to generate the negative voltage.

[tt]On other things, I am still trying to get a quote from David at Cinemag for the transformers :( I have shot him a couple of emails, but it does not seem that after our initial conversation he has had the time to answer it.

Does anybody know another way to contact him? I don't want to be bother too much, but it is kind of one of the critical pieces in the design...

[/tt]

Got him, he had no internet for a while...  he is working in the quote

 

[dmnieto]

I am having an issue with trying to read the values of R370 and R371 in the design... I think they appear to be 10K and 1.43K respectively. With those values the opamp has a zero gain with the EQ enabled and the select on 0dB... But weirdly enough, the +2dB step is actually at +5.4dB according to my simulation.

Does anybody have any suggestion about the values?

Experimentally 10K for R370 and 14K3 for R371 seem to give the right values

 

[dmnieto]

Teaser!

 

[dmnieto]

I wanted to give some signs of life here.

So I finally finished the initial routing and layout of the board. I am going to proceed to the signal integrity analysis and routing optimisation, and of course ensure that the silk-screen is a  bit more pretty than now. Then I will build the prototype.... I am not sure if I will use oshpark or real-pcb. oshpark is my fav proto build, but for the same price I can order 20 boards with real-pcb and they have better lithography.

Just for info, it seems that we are going to be able to knock down the price of the PCB to less that $5 per board... not too shabby!
TOP:

BOTTOM:

 

[dmnieto]

I had a chat with David from Cinemag, he told me that he expects the transformers to be <$75 for both, he needs to custom wind them again from the old specs.

I ordered 2 of each for the prototype, and they should be done in about 3 weeks.

I am working around the cost of the EQ...:

(1) PCB : < $10
(2) Input transformer: ~$50 / (can be bypassed, but then the input gain strap needs to be set to zero to prevent saturating the EQ)
(3) Output transformer: ~$25
(4) Grayhill series 71 switches: ~$20 each (http://www.digikey.com/product-detail/en/71ADF30-02PAJN/GH7328-ND/665931)
(5) Grayhill series 56 switches ~9.50 each (http://www.digikey.com/product-detail/en/56D30-01-1-AJN/GH5601-ND/98646)
(6) Faceplate: ~$50 (as per front panel express, maybe I should do a PCB frontplane?)

So it is trending to ~$253 for the most expensive parts... We should be able to get some discount if we can do a group buy...

 

[kvothe]

And you could put me on the list for a pair, great idea and timing

 

[dmnieto]

And you could put me on the list for a pair, great idea and timing

Talking with David he was like... why don't you add the CA127 preamp as well( he does the input transformer)? I was like, I'd love to... but it is hopelessly impossible to fit
Mybe for the next? CA127 seems like a much easier thing to build

 

[Indecline]

Count me in for a pair. I've been eyeballing the 444 for some time now.

Super surprised you were able to cram it into 500 series format!

What are the gain steps in?

 

[dmnieto]

Count me in for a pair. I've been eyeballing the 444 for some time now.

Super surprised you were able to cram it into 500 series format!

What are the gain steps in?

There are actually no steps on the gain at least in the prototype. It will be +/- 7 or 15dB with a center detent at 0 dB.
The only thing I managed to cram in the design is a EVU-F3M from panasonic (http://www.mouser.com/ds/2/315/P3A1502-197264.pdf) it is a miniature potentiometer.
Unfortunatelly it only comes with a linear taper, so to get the neccesary rev-log I had to come with a circuit trick. I put a resitor in series with the CW and then another parallel resistor with the taper. This gives a very good approximation of a rev-log. and manages to have a very controllable gain and the series resistor allows to callibrate the center detent for equal gain on both of the channels. Even if unfortunately is not stepped.

My wife didn't really matter, as she plans to use the gain to control the stereo width in M/S mode so L/R matching is not a big issue... But I undestand it could be interesting.

It may be possible to get stepped gain by doing the following. Using a rotary switch with an extension of the shaft so it saves the other switches. I have not found any switch yet that does that. But it should be possible. I have placed some anchors in the pcb and the front panel so it will be possible to mount it like that.

Another option that i am going to try is to stick another series 56 switch there, (giving +/- 10dB in 2dB steps) but it is going to be VERY tight. I have made a slot on the PCB to the bottom of the LF switch so it can fit.

Do you think that will be a  better option than the potentiometer?

UPDATE
Thanks Indecine, for pointing me in the right direction... I managed to fit another rotary switch of the 56 series, so now we will have the +/- 10dB in 2dB steps for the output gain. Of course, by changing the resistors you can get whatever you want.

Also, I am in the process of building a prototype PCB front-panel. It will be much cheaper than the metalwork, but also less sexy.

 

[Indecline]

For output gain, I would prefer variable with a zero'd center detent. Having discrete gain steps on the output in mid-side mode may be too much for slight adjustments of the stereo image.. And using smaller steps would limit it's capability in mixing and tracking. Stepped gain in the frequency bands is much more important than in output gain IMO.


Great project BTW!

 

[dmnieto]

Ok,

To be able to use the current structure (which is best for noise), I would need a 9mm rev-log 25K potentiometer with a center detent... Unfortunately I don't think that is available anywhere

The alternative is to use a 10K log potentiometer taper on the positive pad of the AM10, and design the gain of the AM10 to give +10dB of gain end-to-end.

This one coud work.
http://www.mouser.com/ProductDetail/TT-electronics/P081N-271F2CBR10K/?qs=%2fha2pyFadug4XmwqwG0jW5TlQpsAiC01CeQrTyMju671LfzeQaZkQCu1vn%2fY99KV

By putting the CW terminal in series with ~800 and in parallel with 4K5 we get very close to a log taper.

 

[dmnieto]

And this is the final layout of the panel and main PCB. I am working with real-pcb to get the main pcb, the front panel and the AM10 in 2520 in the same panel for a better price... and after this I will send the prototype out.

 

[dmnieto]

Panel completed! Hope to hear from Lynn soon for the final price.

UPDATE: I actually managed to forget that some people may want to run this on 16V rails.
For those there are two options:
- Run the circuit on 16V by jumpering JP22, and then wiring the output transformer as 1:2 to make up for the reduced headroom.
- Use the header JP22 to connect a buck/boost switching generator that provides the full +/- 28V.
This is the layout

Each panel contains 4 opamps, 1 main, 1 front panel and 1 DC-DC converter.

I designed it (based on Matador's booster U87Ai circuit, so props to him) to provide up to 180mA +/- 28V with max 12mV RMS ripple.

It such a small PCB that I don't mind running several iterations to improve it in the future if the proto does not turn out great... but I hope it will.

 

[dmnieto]

Well,
I have an update back for Lynn, and it seems to be more expensive that I expect.

The total quoted price (including shipping) for the panel is $375 for 20 panels,  this comes for approximately $18.78 ($20.15 ENIG) per set of a single channel PCB. It has a ~$212 tooling cost. So subsequent runs will only be at about $13.18 ($14.55 with ENIG finish).

The issue is the minimum cost with them. For $375 I can afford two-respins of the main board with oshpark, so I am going to go the route of ordering a 3 board set with osh-park. And when it is totally proven, I will build a pair, and share how it sounds here. I want to be open here, in that I would like to recover a bit the cost of the design, so I will probably go for either a HASL or ENIG finish and offer the PCB for about $25 per channel to the first orderers.

From there, I am not sure how to proceed honestly. It is my intention to offer this design for free, or at a minimum operating benefit and donate the proceeds to groupdiy. If I go with real-pcb the tooling cost would disappear and the cost of $13-$15 per PCB would be available to anybody to order.

I don't have the time to manage the ordering of the boards, but if somebody (like gustav, hairball, chunger, or other store owners) would agree to split 50/50 with groupdiy any benefits of the sale of any future boards I would gladly give the design to them.

Anybody can chip in some ideas? What I want to avoid is situations like microphone-parts in which they would stole away the design with no profit for the community.

Thanks!

 

[dmnieto]

David from Cinemag has finished winding the QEE transformer prototypes and they are on its way. They came at slightly over the budget (because of being prototypes and low volume) at $77.08 for both of them.

The most expensive is the input transformer, as it has some special lamination topology, but as Ken said in this orphan audio forum http://www.orphanaudio.com/forums/viewtopic.php?f=8&t=2475#p4340

All "stock" 333c,s have a 4:1 bridging input transformer with the first gain stage modified to make up the level, and the final amp stage has a 1k rev-log level pot to give +/-15db of gain adjust. ("stock" since the 333c was custom built for only two consoles, Lions Gate films and The Cary Grant dub theatre at what is now Sony pictures). The input transformer is completely responsible for the sonics of this version of the eq and the 30 step precision pots on frequency select and gain/cut make it one of the most versatile and flexible eq,s of its type. Add the continuous swept HI/Low pass filter pots and you have an eq that is very hard to beat for twice its value (which is a lot more than you managed to get your 333c for).

For budget configurations, the sockets in the PCB are compatible with Edcors XSM series.
For the output transformers, both the Edcor and the QEE have steel laminations, but the QEE has higher primary inductance and consequently lower THD. Also, the QEE can be driven hotter.
For the input transformers, the QEE is 80% nickel... vs the Edcor which is steel. I expect the sound to be quite different.

I am also considering adding support in the PCB for the JT-10KB-DPC, but it is rated to only 21dB vs the 30dB of the QEE.
For the output, there are plenty of Jensen transformer that will work:
JT-11SS-DLCF (1:1 +30dBu) (recommended)
JT-11-DMCF(1:1 +22dBu)



I expect the PCBs to be done in a couple of days (plus shipping), so it is going to be an interesting week of testing

 

[dmnieto]

I received the PCBs and the transformers today.

And the initial feedback is that I did not mess the mechanical!

 

[dmnieto]

testing the transformers

Good square response, may need some dampening

 

[dmnieto]

Color me impressed, wow, actually wow.

I have tested the QEE-3041 output transformer that David from Cinemag sent me. Given its relative small size (same as http://cinemag.biz/output/PDF/CMOB-2.pdf) the f*cker is delivering +32dBu at 20Khz, with a 600Ohm load in a 1:2 step-up configuration (Rs 150Ohm) with less than 0.6%THD.

I can wait to listen to them

And they are just $25 each

 

[dmnieto]

I just need the rest of the pieces to start building!