I changed the input filter to 33pF. Nothing really changed for the OPA177. Still garbage. Cf411 works well up to 500khz, then really loses speed and gain. LT1028 runs perfect up to 1gHz. That's the limit of my signal generator. My LT1001s came in as SMT. I've got OPA177 in the Jungs for now. I'll test some more when the proper ones arrive on Tuesday.
I should mention the Jungs have no measurable noise until the the linestage is clipping and the output is 15V. At 4V output it's perfect. Gain is around 6x. I need better test equipment though. My scope is crap and my good meter is out for calibration right now.
I should mention the Jungs have no measurable noise until the the linestage is clipping and the output is 15V. At 4V output it's perfect. Gain is around 6x. I need better test equipment though. My scope is crap and my good meter is out for calibration right now.
So we have Jungs with -120db PSRR and no measurable noise. HAH !
All those IC op-amps at least run .... no special parts ... good.
These regulators are quite likely noisier than the original Jung. It bypassed
the error-amp with a cap to feed a noise correction signal back to the
pass device.
But , they "smoke" a 317 ..... and the added linestage PSRR and noise rejection
will negate a "pitchfork Jung".
This mean a Jung with the shunt added will most likely beat the original
(for the digitals).
We can't expect any op-amp for the linestage to give top performance ,
if a particular one gives superior slew/bandwidth - go with it.
PS- 75ma class A is about ideal. I simulated at 80.
Edit - the OPA177 is a low bandwidth BJT !! - general purpose IC ... better for regulator stability.
The others are real audio devices .... just looked the datasheets up.
OS
Last edited:
Do you think the Jung pass transistors will run cooler with a lower voltage transformer? I was testing with 22VAC. 12VAC ran cold but didn't quite make 15V.
Don't the pass devices mount below on the chassis ? They should have no
issue with a 12V drop and <1/4A dissipation.
OS
Right now I have them mounted on a 5" x 6" piece of aluminum plate below the board. They get the plate up to 130 degrees after about 10 minutes. It's likely going to pump out more heat after the volume pot and analogue parts are connected. Not dangerously hot but it could run cooler.
130 C or F ? If it's F , 5 X 6" is a low C/W rating. That protection board
7812 will warm up that whole 7 X 9" plate you made me with only >100ma.
The regulator transistors should dissipate 3-4W each.
OS
130F. It's not really a concern unless I mount this up on a backplane and can't get these transistors against the chassis.
Linestage Porn!😱
I ran it a little more with a 10k load on the output. It definitely needs more heat sink on the outputs but I was expecting that. 22pF on the input filter would probably be good. Output is still starting to drop at 500MHz with 33pF.
I ran it a little more with a 10k load on the output. It definitely needs more heat sink on the outputs but I was expecting that. 22pF on the input filter would probably be good. Output is still starting to drop at 500MHz with 33pF.
Attachments
Sweet , looks better than many a Ebay kit !
You could drop to 330R for those linestage CCS's (cooler - 54ma class A).
Load on the output is very little of the dissipation , the class A iis most
of it (even down to <100R load).
OS
Do you think we should have multiple digital inputs? I run everything through my PC but an input from a blu-ray or game console might be an option.
Of course, suggest use the wm8805 instead. You have 8 channels to use. Then you have to consider, you mix of input types, do you want to use an s/pdif isolation transformer perhaps on one channel or more.
The s/w for both wm8804/5 is very similar except the xtra reg's for channel selection/status.
also suggest to first time power up in h/w mode and then when you get your s/w figured out, re-configure reset strapping for s/w mode to get all your h/w features going. having the mcu do it, you can make it dynamically configurable in the code. this is theory until tested!!
What is your plan jeff, make all your boards <300 pins and use the free diptrace s/w?
I was having dip trace crash on me running on my HP laptop w/vista o/s? But then again so does orcad 16 🙂
The s/w for both wm8804/5 is very similar except the xtra reg's for channel selection/status.
also suggest to first time power up in h/w mode and then when you get your s/w figured out, re-configure reset strapping for s/w mode to get all your h/w features going. having the mcu do it, you can make it dynamically configurable in the code. this is theory until tested!!
What is your plan jeff, make all your boards <300 pins and use the free diptrace s/w?
I was having dip trace crash on me running on my HP laptop w/vista o/s? But then again so does orcad 16 🙂
I've been looking at WM8805 and CS8416. The CS8416 is going obsolete but can select between 4 inputs in hardware mode. All coax inputs will through pulse transformers. I want to eliminate any chance of ground loops or supply noise I can.
I'm running Diptrace Lite 500 pin free version right (free upgrade if you email them and tell them it's for hobby) now but will upgrade the licence on one of my pcs if needed.
I'm going to either hand etch a tester board or get some modular tester boards produced soon. I like to breadboard projects before I build a finished product. I seem to always get the tricky stuff right the first time but overlook the obvious. Always a cap or diode in backwards somewhere.
I think just about everything including Vista itself will cause that POS operating system to crash. I ran it for 2 days when it came out and upgraded to XP.
Well i did give you my cs8416 design to work from if you go that route.
I guess that is one advantage of the cs8416, but in the long run you probably want to get the s/w mode going anyways.
The reason that cs8416 is obsolete is because to me, the wm8805 is a better part, it has the master clock gen built-in, it can operate in both master and slave i2s modes has a transmitter for loop thru/back. It is a more complicated beast to program from what I have read so far but is well documented.
Hand etch for TSSOP? what pin pitch can you etch? 0.5,0.65,1mm? You really should do a double sided pcb with a solid ground plane to get best performance.
thx for the dip trace hint to upgrade to <500 pin version. Still it is a hard limit to work around.
I will try diptrace on my pc running xp and see if it is for sure the vista pos. I do not want to give MS anymore $.
Sure it is always something that gets you, in a pcb design, always good too have more than one set of eyes looking over a design.
I can look over your designs assuming that I can review what you have done. I use gerbtool for gerber/drill/cnc data.
Last edited:
That picture I posted of Calvin with an Atmega128 board was an actual board I used to hand etch years ago. With good clean 1/2 ounce copper anything single or double sided can be hand etched with the TTS method if you don't cut corners. I have a dedicated laminator just for this. The biggest limiting factor is the different expansion rates for the board and the TTS paper. The paper cools faster than the board and pulls the toner off on larger boards. I had good experiences with photo-fab as well except I always had problems with some idiot coming into the room and turning on the lights for me so I gave up on that.
Everything will likely work better in software mode but I like the idea of keeping everything as simple as possible with as many hardware mode options as possible. There's no reason why a Peamp / DAC can't be operated from a rotary input selector switch and a stereo pot if one chooses to do so. It's easy enough to add a couple jumpers or dip switches to select software / hardware modes when assembling.
A second set of eyes is always good. After staring at a board for too long everything becomes a blur. That's why I switched to Diptrace. I let it verify that the schematic and board match. I always try to post schematics hoping someone will look at them closely and catch any mistakes. It hasn't been working out so well though. Nobody caught the inverted diodes in the VAS section of the linestages.
Sure I agree with all that you said with stuffing options. I have seen a stereo deigned from 4000 cmos logic/state machines, to simple mcu's. It comes down to the feature set, ease of use, UI.
Looks as if you still have not nailed down your feature set yet. Designing to a moving target is exhausting.
Like do you want remote control or not? changes the game entirely.
With smt you can pack a ton of stuff in a very small space, so I say to make the most of it.
Sorry I did not review/sim your posted design.
For first time success, it helps a lot to have good tools and libraries, all interconnected. Using sim as part of the design process helps, as much as drc rules do. When you add the s/w or f/w component, it becomes even more difficult to sim.
Mine's getting a simple remote and display. I'm trying to design it with options for anyone else that may want to build it. If it's designed to be able to run in hardware mode it's simple to add a couple relays to operate from a basic remote and add some data busses for more advanced interface. It's tougher to go from an advanced control design to simple controls.
OS mentioned he would like to keep as much as possible in through hole to make it easier for most builders. I've been trying to keep everything through hole or large two row surface mount. Four sided is miserable to rework if needed.
Sounds good, lets see what you guys come up with.
So far i read,
PitchFork feature set.
1) spdif receiver
2) stereo dac
3) stereo analog source selector
4) stereo volume control
5) stereo headphone, line amp
6) Control/display
7) power supply
All in one box/size?
What music source do you mostly listen too?
So far i read,
PitchFork feature set.
1) spdif receiver
2) stereo dac
3) stereo analog source selector
4) stereo volume control
5) stereo headphone, line amp
6) Control/display
7) power supply
All in one box/size?
What music source do you mostly listen too?
OS mentioned he preferred toslink for noise immunity. The one I'm building right now runs from a pc only.
My theater system is going to need multiple analogue and digital inputs. HDMI and likely display port too. That one's going to need an advanced interface. This project is starting point/test bench for that one for me.
SPDIF can also be isolated with a small transformer.
My asus MB only has toslink as digital out.
Older MB's either are spdif/toslink or just coax spdif.
So , the newer MB's seem to be moving to all toslink (realtech alc887 codec).
OS