goldenreference low voltage power supply

[jose]

I´m sure that Opa is OK,I have changed it three or four times. Sorenb, yes I measure -3.23volt across R9 and 10V across R10. 

I was work on this área and I changed R9, R10, 1N914 diodes and C3. I´m not sure but I found this: 

 

Probably has nothing to do but who knows...

I measured C3 and it´s  9,8uF but I changed it for other. 

Finally:

 

 

Thanks for your help and patience boys

 

[sorenb]

the resistor you are holding in your hand, is that either R9 or R10 then or?

[jose]

R9. When I removed the resistance I kept a piece of it in the pliers.  

I'm not sure if this is the cause of the problem or it was the tantalum condenser. The Zeners were fine and R10 also seemed to be correct. 

[Pars]

Hmm. Opening up R9 in a sim gives ~24V with a lot of ripple. Opening R10 gives something similar to what you were seeing, output at ~11Vdc (clean).

[Pars]

14 hours ago, ruslanch said:

I remade my previous GRLV pcb for my amplifier:
1. Optimized the pcb size.
2. Fixed bugs of the previous version - added forgotten capacitors in the supply chain for op amps, the supply of all internal circuits takes from the pure part of the circuit, as in the original version of the board.
3. Now this is the native KiCad project, not the conversion from Eagle.

Gerbers, KiCad project, 3D view 

Nice job! KiCad perplexes me, steep learning curve.

[ruslanch]

14 hours ago, Pars said:

Nice job! KiCad perplexes me, steep learning curve.

Thanks! Initially, I began making a scheme in Eagle CAD and it turned out to be very convenient. But when I started making the PCB layout, then rested in the restrictions of the free version of Eagle, and continued the layout in KiCad. In fact, the difference is small, especially if in KiCad enable the OpenGL mode. The interactive router in KiCad is a bit blunt, it needs a little more tips, but this is a very small drawback. For a new version of the PCB layout, I completely remade it from scratch in KiCad, and now I find KiCad very convenient.

PS. Has slightly updated the project - added a fill of the bottom layer and added 10 ohm resistor for grounds separation.

[Pars]

I had tried doing some mods to your layout in KiCad (going back to the RS402 bridges, and adding a Phoenix 508/Molex KK jack that I did in an Eagle library). I was having a hard time figuring out the libraries in KiCad and the schematic image vs. footprint.
I’m using a Mac, so not sure the OpenGL is even an option in that...


Sent from my iPhone using Tapatalk

[gepardcv]

@ruslanch, @Pars: How large is the redone GRLV board?

[Pars]

From what I recall when viewing the gerbers, it was around 106 x 96? Larger than I can do in my free Eagle copy.
I’ve taken to using Diptrace lately for a board I sent off for production that was 142x95. It seems easier to pick up for me than KiCad, but I’ll keep looking at it.
IMO, Eagle is shooting themselves in the foot with their pricing, particularly since Autodesk bought them. Integration with Fusion360, while in the “gee neato” category, isn’t useful enough to make me cough up for it. Others mileage may vary.


Sent from my iPhone using Tapatalk

[jose]

20 hours ago, Pars said:

Hmm. Opening up R9 in a sim gives ~24V with a lot of ripple. Opening R10 gives something similar to what you were seeing, output at ~11Vdc (clean).

The side that broke is the one that is in contact with R10. Could this be the cause?

[ruslanch]

17 hours ago, gepardcv said:

@ruslanch, @Pars: How large is the redone GRLV board?

129x99mm

17 hours ago, Pars said:

I had tried doing some mods to your layout in KiCad (going back to the RS402 bridges, and adding a Phoenix 508/Molex KK jack that I did in an Eagle library). I was having a hard time figuring out the libraries in KiCad and the schematic image vs. footprint.
I’m using a Mac, so not sure the OpenGL is even an option in that...

Something like this: Gerbers, KiCad project, 3D view. Yes, finding the components in the library is sometimes difficult ).

[masamoto]

On 19.12.2017 at 6:35 PM, cspirou said:

With 24vac you are getting 33Vdc before regulation. Thats a 13v drop and far more then you need. The problem though is for a given VA rating the current capacity goes down as the voltage goes up. You probably sourced too much current and burned out the windings.

get 2 x 18v instead 

With much less to drop the MJW's are a lot cooler as well. And the transformer was not even warm to the touch.

Thanks guys. So much more to learn

[Pars]

5 hours ago, jose said:

The side that broke is the one that is in contact with R10. Could this be the cause?

Not really in contact with R10, so no, I don't think so. Of course sims aren't completely accurate either.

[kevin gilmore]

because soren is running the grlv supply at 4.5 amperes, here is a new version with the feedback resistor locations flipped so that the sense is not one inch away from the output connector.

sense is now tied directly to the inner power supply connector. So the 20mv change in voltage due to current should be completely canceled out.

for higher currents, best if the board is made with 3 oz copper.

 

goldenreference6drflip.zip

Edited January 3, 2018 by kevin gilmore

[sorenb]

On 4/23/2016 at 1:40 PM, kevin gilmore said:

goldenreference is only good for about 1.5 amps

when you need a lot more than that, you have to add an extra set of pass transistors. like the ones on the uber2amp board. And they have to be mounted to a large heatsink

did a load test of a stock GRLV configured for +/-20V, using a lab supply as source . 
Unloaded 20.0556V out, and at 4.25A it drops to 20.0333V - caused by the resistance of the trace (between sensing resistor and terminal block) itself.
Seems to be stable regulating as long as the voltage drop across the pass transistor is > 3V (Kevin recommends > 4V though).
Also going above 1.5A one might benefit from upping the current to the driver transistor (Kevin mention this earlier in this thread) and using larger caps.
The largest I can find that fits is this https://www.mouser.dk/Passive-Components/Capacitors/Aluminum-Electrolytic-Capacitors/Aluminum-Electrolytic-Capacitors-Snap-In/_/N-75hr0?P=1z0j7l0Z1z0x3saZ1yx4awyZ1yx4atvZ1ypb2ikZ1z0vn3sZ1z0wqdeZ1z0wqf4

 

Using a 24VAC 100VA transformer, I get almost the same.
At 4.2A the regulation drops at 4.2V Vce (21.6VAC input).
Seems like > 4V (as Kevin has stated earlier) is what is need when < 1.5A, and more like 7v if approaching 4A (and raising the current for the driver)

 

Edited January 3, 2018 by sorenb

[cspirou]

According to the datasheet for the MJW21194, they are rated up to 16A. I wonder how far the GRLV can be pushed. Could it power a First Watt F5?

[mwl168]

2 hours ago, sorenb said:

did a load test of a stock GRLV configured for +/-20V, using a lab supply as source . 
Unloaded there is 20.0556V out, and at 4.25A it drops to 20.0333V - caused by the resistance of the trace (between sensing resistor and terminal block) itself.
Seems to be stable regulating as long as the voltage drop across the pass transistor is > 3V (Kevin recommends > 4V though).
Also going above 1.5A one might benefit from upping the current to the driver transistor (Kevin mention this earlier in this thread) and using larger caps.
The largest I can find that fits is this https://www.mouser.dk/Passive-Components/Capacitors/Aluminum-Electrolytic-Capacitors/Aluminum-Electrolytic-Capacitors-Snap-In/_/N-75hr0?P=1z0j7l0Z1z0x3saZ1yx4awyZ1yx4atvZ1ypb2ikZ1z0vn3sZ1z0wqdeZ1z0wqf4

 

Have not measured the voltage drop on the pass transistor myself, but how does this 3V or 4V drop translate to the raw DC voltage at the rectifier?

Drawing about 1A per rail, my GRLV could not maintain regulation for 24VDC rails when supplied by a transformer with 24vac/2A secondaries.

 

[sorenb]

2 hours ago, mwl168 said:

Drawing about 1A per rail, my GRLV could not maintain regulation for 24VDC rails when supplied by a transformer with 24vac/2A secondaries.

"could not maintain regulation" meaning what exactly?

(edit after adding the transformer info to the previousposting)
You probably want to check the voltages in the input pins of the OpAmp (directly on the OpAmp) against the pin6 at the LT1021, when you experience the regulation doesn't work

Edited January 3, 2018 by sorenb

[kevin gilmore]

here is the version with the to220 sic diodes, someone really needs to check this

 

 

goldenreference6drflipto220.zip

Edited January 4, 2018 by kevin gilmore

[mypasswordis]

On 1/2/2018 at 2:20 PM, ruslanch said:

Something like this: Gerbers, KiCad project, 3D view. Yes, finding the components in the library is sometimes difficult ).

Definitely the worst part of Kicad is the libraries. One thing I learned the hard way (arcy sparky blowup) is don't assume the pins of the footprints are what you think they are, or even that they will stay consistent, or just make your own components and footprints. It is improving a lot recently from CERN intervention. 

I need to revisit Eagle as the integration into Fusion is very cool, but from what I remember it was very hard to use. 

[luvdunhill]

According to the datasheet for the MJW21194, they are rated up to 16A. I wonder how far the GRLV can be pushed. Could it power a First Watt F5?

Why? I can get less than -100dB line rejection (60Hz and higher harmonics) from a properly designed snubber, careful layout .. with a CRC supply. Now, it took some care.. but. I do have PCBs for sale motmotmotmot.

[cspirou]

27 minutes ago, luvdunhill said:

Why?

 

Just curiosity as to how universal the GRLV is.

28 minutes ago, luvdunhill said:

 I do have PCBs for sale motmotmotmot.

 

Interested. 

[Pars]

3 hours ago, kevin gilmore said:

here is the version with the to220 sic diodes, someone really needs to check this

goldenreference6drflipto220.zip

Looks OK from what I looked at. Didn't do a full compare.

Trace clearances are tight; the trace running down the left side of the right bank of diodes (when viewed text correct way up) is tight on the pads; same with the trace running between the bottom 2 rows of diodes in the left 2 banks.

How big in diameter are the two caps? 35mm?

EDIT: Needs MOAR outputs (make the board bigger... still small at 100 x 110

Edited January 3, 2018 by Pars

[kevin gilmore]

updated the file.

the caps are 25mm

 

[mwl168]

6 hours ago, sorenb said:

"could not maintain regulation" meaning what exactly?

 

Fed by a transformer with 2 X 24vac/2A secondaries, my GRLV was not able to sustain regulation for 24 VDC output when the current draw was about 1A per rail. The output sagged down to about 23 VDC.  

The same setup works fine when I reduce the current draw to about 500mA pre rail. It also works fine for 1 A per rail draw once I lowered the regulated output to 22 VDC.