|
Post by indiehouse on Nov 10, 2014 15:10:30 GMT -6
All you have to do is do it in 6 months and you're ahead of the game 6 months? Aren't we up to 7 and counting now?
|
|
|
Post by mdmitch2 on Nov 10, 2014 15:24:55 GMT -6
Shoot, I'm aiming for 6 weeks. 6 weeks! Sounds like a nice Christmas present
|
|
|
Post by tonycamphd on Nov 10, 2014 15:47:44 GMT -6
i think dandeurlou and cowboy are moving into the 9 month range!
|
|
|
Post by wiz on Nov 10, 2014 17:27:19 GMT -6
I dont understand half of what you wrote svartbut gee I am glad you wrote it bud. great stuff just remember Oz is 240V 50Hz 8) cheers Wiz
|
|
ericn
Temp
Balance Engineer
Posts: 16,103
|
Post by ericn on Nov 10, 2014 17:47:21 GMT -6
Shoot, I'm aiming for 6 weeks. Hey , I was thinking of pulling the trigger on a RMA knowing it might be 2016 before I saw it , Then you strayed this project ! I'm in once you get everything down and hey I waited over a year for my M49 bodies!
|
|
|
Post by dandeurloo on Nov 11, 2014 9:26:20 GMT -6
I'm still waiting for RMA and m49 bodies. Dang!
|
|
|
Post by jimwilliams on Nov 11, 2014 10:37:16 GMT -6
I wouldn't use the LM4562 for the ADC buffers, not good enough. Avoid the PGA/INA Burrbrown part used on the demo pcb, it sounds cloudy. Many others have also complained about that front end design. I used the opamp based design shown in the PCM4222 data sheets, the non-inverting version. I set the feedback resistor to 1k ohms to obtain +19 dbu at zero. I used the excellent Analog devices ADA4898-2 dual opamp at .9 nv noise, 60 v/us slew rate and very low THD. I had Ross Martin make my DAC with those opamps. They apparently impressed him so much he now offers them as an option. Ok, I'll give them a test as well, but it'd be hard to beat Jung's swan song. I switched the design to the alternate datasheet inverting opamp config and so they are also a standard SOIC8 footprint now. This allows more options for change as testing ensues on the prototypes, whenever that happens. The non-inverting configuration offers a selectable input impedance and a lower noise design as the series resistor in the inverting configuration adds additional resistor noise. That will add to the .9 nv noise spec of the ADA4898-2 opamp if that resistor is above about 200 ohms. The idea of using the AD chip is to set the noise floor below the converter dynamic range. The ADA4898-2 is a high current part, about 8 ma per 1/2. It also runs very hot as it has a close to a class A output stage. There is a steel paddle on the underside intended to be soldered onto a few vias to heat sink the part. If you layout for that part, be aware of the thermal issues and provide the vias for heatsinking. If you use the single package version you won't have those thermal issues and you will also have more options for opamp selection. The ADA4627-1 is another excellent sounding opamp. fet input with over a 100v/us slew rate.
|
|
|
Post by svart on Nov 11, 2014 10:57:14 GMT -6
Ok, I'll give them a test as well, but it'd be hard to beat Jung's swan song. I switched the design to the alternate datasheet inverting opamp config and so they are also a standard SOIC8 footprint now. This allows more options for change as testing ensues on the prototypes, whenever that happens. The non-inverting configuration offers a selectable input impedance and a lower noise design as the series resistor in the inverting configuration adds additional resistor noise. That will add to the .9 nv noise spec of the ADA4898-2 opamp if that resistor is above about 200 ohms. The idea of using the AD chip is to set the noise floor below the converter dynamic range. The ADA4898-2 is a high current part, about 8 ma per 1/2. It also runs very hot as it has a close to a class A output stage. There is a steel paddle on the underside intended to be soldered onto a few vias to heat sink the part. If you layout for that part, be aware of the thermal issues and provide the vias for heatsinking. If you use the single package version you won't have those thermal issues and you will also have more options for opamp selection. The ADA4627-1 is another excellent sounding opamp. fet input with over a 100v/us slew rate. Understood. However, as the other Jim Williams used to say, "always invert, unless you can't". This was primarily from the viewpoint of higher precision, less issues with resistor noise in the feedback and much better CMRR(which in mixed signal designs like this is a bonus), but as you mention, it's rather academic in such a project as I'll have much more to worry about with noise ingresses due to layout density. Due to density I'll probably stay with dual opamps on the frontend, but I'll add a ground paddle underneath for those opamps that need it.
|
|
|
Post by svart on Nov 11, 2014 14:14:36 GMT -6
I'm trying very hard to do this in 2 layer, but keeping ground integrity as well as route integrity is becoming very hard.
This might have to be a 4 layer board after all.
|
|
|
Post by tonycamphd on Nov 11, 2014 14:28:48 GMT -6
I'm trying very hard to do this in 2 layer, but keeping ground integrity as well as route integrity is becoming very hard. This might have to be a 4 layer board after all. what ever doesn't compromise the quality gets my vote, what is the cost diff between 2/4 layers generally?
|
|
|
Post by svart on Nov 11, 2014 14:31:45 GMT -6
In small quantities of 25 or less, about 10$ more per PCB. It becomes cheaper as you go up in quantity of course. That's an estimate based on automated online quoting.
|
|
|
Post by Johnkenn on Nov 11, 2014 16:08:56 GMT -6
I'm trying very hard to do this in 2 layer, but keeping ground integrity as well as route integrity is becoming very hard. This might have to be a 4 layer board after all. what ever doesn't compromise the quality gets my vote, what is the cost diff between 2/4 layers generally? There are two additional layers...
|
|
|
Post by svart on Nov 11, 2014 16:21:41 GMT -6
what ever doesn't compromise the quality gets my vote, what is the cost diff between 2/4 layers generally? There are two additional layers... I'm unsure of what the annoyed smiley is about..
|
|
|
Post by Johnkenn on Nov 11, 2014 16:30:45 GMT -6
I was trying to be funny...but I misread it...I thought he asked, "What is the difference between 2/4 layers?" I thought I would be funny...but alas I am not...and I am a fool.
|
|
|
Post by wiz on Nov 11, 2014 17:10:14 GMT -6
I'm trying very hard to do this in 2 layer, but keeping ground integrity as well as route integrity is becoming very hard. This might have to be a 4 layer board after all. NURSE.. .wipe that mans brow... bring him some OJ and someone give this man a back rub ...STAT now, SVART GET BACK TO WORK!!! 8) cheers Wiz
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 11, 2014 20:56:00 GMT -6
:DI thought John wanted to use "the force" on svart in a hypnotic Jedi manner. LOL
Alec Guiness-voice: "You want to do uncompromised quality." Svart: "I want to do uncompromised quality." Alec Guiness-voice: "The board will have 4 layers." Svart: "I will do 4 layers. I go back to work..."
|
|
|
Post by svart on Nov 12, 2014 9:41:00 GMT -6
Teaser..
|
|
|
Post by tonycamphd on Nov 12, 2014 10:02:01 GMT -6
What is that? U couldn't have possibly burned a board that fast!?
|
|
|
Post by mdmitch2 on Nov 12, 2014 10:04:45 GMT -6
Fancy pants rendering. Neato.
|
|
|
Post by svart on Nov 12, 2014 10:12:23 GMT -6
What is that? U couldn't have possibly burned a board that fast!? Nah, it's a 3D model of the PCB so far. I still have a lot to do, and probably optimization of traces after that. The PCB would take about a week for the first ones to return from the board house. I also got confirmation from the oscillator company that my samples have shipped. Once I send the prototype boards out, I'll order enough parts for 3 prototypes. Once I confirm that the oscillators work well, I'll order enough for a larger run. The oscillators are custom and have a 2-3 week lead time for small quantities. I still have a bunch of code to do as well, but I think I'll wait til the boards are sent and use the down time to work on it. I bought another small Arduino board so I can work on the communications between controllers as well. Also, I can't seem to find a viable option for an external power supply. There seems to be everything but what I would need. I suppose this means I need to go with an internal supply.
|
|
|
Post by jimwilliams on Nov 12, 2014 10:50:22 GMT -6
Internal should be ok if you use a torrid transformer and wrap a piece of steel around it. I got 60/120 hz down to -125 dbu that way using big caps, before and after the regulators. Low impedance, low dropout regulators like the LT1086 are the best integrated solutions, discrete regs are even better, check out the Diddon/Jung designs.
|
|
|
Post by tonycamphd on Nov 12, 2014 10:52:49 GMT -6
What is that? U couldn't have possibly burned a board that fast!? Nah, it's a 3D model of the PCB so far. I still have a lot to do, and probably optimization of traces after that. The PCB would take about a week for the first ones to return from the board house. I also got confirmation from the oscillator company that my samples have shipped. Once I send the prototype boards out, I'll order enough parts for 3 prototypes. Once I confirm that the oscillators work well, I'll order enough for a larger run. The oscillators are custom and have a 2-3 week lead time for small quantities. I still have a bunch of code to do as well, but I think I'll wait til the boards are sent and use the down time to work on it. I bought another small Arduino board so I can work on the communications between controllers as well. Also, I can't seem to find a viable option for an external power supply. There seems to be everything but what I would need. I suppose this means I need to go with an internal supply. Lol, I was looking on my phone... Won't an internal ps cause cost prohibitive problems?
|
|
|
Post by svart on Nov 12, 2014 12:26:15 GMT -6
Nah, it's a 3D model of the PCB so far. I still have a lot to do, and probably optimization of traces after that. The PCB would take about a week for the first ones to return from the board house. I also got confirmation from the oscillator company that my samples have shipped. Once I send the prototype boards out, I'll order enough parts for 3 prototypes. Once I confirm that the oscillators work well, I'll order enough for a larger run. The oscillators are custom and have a 2-3 week lead time for small quantities. I still have a bunch of code to do as well, but I think I'll wait til the boards are sent and use the down time to work on it. I bought another small Arduino board so I can work on the communications between controllers as well. Also, I can't seem to find a viable option for an external power supply. There seems to be everything but what I would need. I suppose this means I need to go with an internal supply. Lol, I was looking on my phone... Won't an internal ps cause cost prohibitive problems? Not necessarily with the design integrity aspects, just cost and complexity. A nice toroid tranny will be fine depending on positioning and shielding. I was thinking more of the costs associated with adding these things. Now there will need to be 120/240 switching, fusing, AC receptacle, hand wiring, etc. Not to mention more complex metalwork. Secondly, I need to do more research on FCC/CE/UL compliance testing for internal AC supplies. In the places I've worked, we've always worked hard to stay with external supplies that are already FCC/CE/UL compliant so that we could skip a lot of testing ($$$$$). I'm not sure if it matters for small quantities, and UL is not a federal mandate like CE is.
|
|
|
Post by svart on Nov 12, 2014 12:33:46 GMT -6
Internal should be ok if you use a torrid transformer and wrap a piece of steel around it. I got 60/120 hz down to -125 dbu that way using big caps, before and after the regulators. Low impedance, low dropout regulators like the LT1086 are the best integrated solutions, discrete regs are even better, check out the Diddon/Jung designs. I've used the LT1086 before, as well as some others, but the others were cost prohibitive. I've chosen the TPS7A70xx and TPS7A33xx series LDOs from TI for the +/-15V on this one though. Roughly similar specs but in a smaller package. I've not used the TI regs before, so we'll see.
|
|
|
Post by wiz on Nov 12, 2014 16:07:12 GMT -6
Internal should be ok if you use a torrid transformer and wrap a piece of steel around it. I got 60/120 hz down to -125 dbu that way using big caps, before and after the regulators. Low impedance, low dropout regulators like the LT1086 are the best integrated solutions, discrete regs are even better, check out the Diddon/Jung designs. Goss Band...I had to fit one on the toroidal in my 1073DMP cheers Wiz
|
|