Hey,
I'm new to these improved psu designs but would like to try some for +30/-10V operation. Current load shouldnt go over the rated 1.5A max of the LM317 so from the published ckts I assume I don't need to change anything except the feedback resistors around the opamp to set the output voltage...
Here's a schemo I've drawn up of an exsisting schem found on this forum somewhere (of the + supply):
I've layed out a single sided PCB too: (inspired from designs whilst searching this forum):
We good to roll?
Cheers,
I'm new to these improved psu designs but would like to try some for +30/-10V operation. Current load shouldnt go over the rated 1.5A max of the LM317 so from the published ckts I assume I don't need to change anything except the feedback resistors around the opamp to set the output voltage...
Here's a schemo I've drawn up of an exsisting schem found on this forum somewhere (of the + supply):
An externally hosted image should be here but it was not working when we last tested it.
I've layed out a single sided PCB too: (inspired from designs whilst searching this forum):
An externally hosted image should be here but it was not working when we last tested it.
We good to roll?
Cheers,
daArry said:
Yeah, looks not bad.
Couple points (in the positive spirit): C2 is a waste. The diode will have a dynamic impedance of say 1 Ohm. Plus the intrinsic emitter resistance that is unbypassed. That 100uF will not measurable change things, let alone audible. It takes up too much PCB space.
The PCB can be improved.
At the left, the area C1, C9, C8 can be MUCH tighter, making for shorter connections. Rotate caps for shortest tracks. On the right side, R7, ZD1, RV2, R5 should be moved directly close to the opamp. C4 can be rotated, all to give a smaller, tighter PCB. Don't make wide tracks for small-signal lines. they take up too much space. Wide tracks here only make sense for the input, output and ground lines. The rest is all small-signal.
But your layout is clean and the grounding looks OK to me.
Jan Didden
Hi Jan,
Thanks for your comments - here's an updated layout:
Much smaller brd size now
All the best,
Dan
Thanks for your comments - here's an updated layout:
An externally hosted image should be here but it was not working when we last tested it.
Much smaller brd size now
All the best,
Dan
daArry said:
Thanks for your comments - here's an updated layout:
An externally hosted image should be here but it was not working when we last tested it.
Much smaller brd size now
The board looks great now.
You could accomodate things a bit more by moving Vout out to the border of the pcb and GNDout to be just below it. As it is now it's difficult to get to.
If you rotate C5 clockwise and move it up a bit you might be able to do that.
On the lower side C6 could be rotated to be 90 degrees to the IC, allowing a bit more space for different types.
Include AD817 as a DIP option to the AD797, which seems to be a "conflictive" option.
If you are considering an only-SOIC part as the AD825 you should put the tracks for it on the pcb, if you are designing it from zero.
One of the good things of the Jung regulator was the "sensing" ground. Aren't you implementing any?
Carlos
Hey Carlos,
The reason for the placement of the vin/vout, gnd_in/out holes was so I could sit the pcb on vero stripboard and each would have it's own copper line from the off. Then, on one end I'd have the rectifying ckt and the other the regulated outs on screwfix connectors...
Thanks for the heads-up on the opamps - for the smd only parts i was gonna attach the part to a dip ic socket by bending the legs outwards and soldering it underneath...
WRT the sensing ground - from what i've seen it's all joined up somewhere along the way anyway so I was just goin for a quicker solution by attaching all grounds on the brd from the get go...or have i missed something?
The reason for the placement of the vin/vout, gnd_in/out holes was so I could sit the pcb on vero stripboard and each would have it's own copper line from the off. Then, on one end I'd have the rectifying ckt and the other the regulated outs on screwfix connectors...
Thanks for the heads-up on the opamps - for the smd only parts i was gonna attach the part to a dip ic socket by bending the legs outwards and soldering it underneath...
WRT the sensing ground - from what i've seen it's all joined up somewhere along the way anyway so I was just goin for a quicker solution by attaching all grounds on the brd from the get go...or have i missed something?
daArry said:
Well, you missed why the grounds should be kept separate and only joined at the powering point, close to the device you are going to power.
Perhaps you should read the original Gary Galo article on the matter.
Whether this issue is important and how, it would be nice to hear from others too.
Carlos
Yer, I (try n') aim for a star/central point system when grounding so I guess might as well here then eh
Here's a tweaked layout with the 0Vsense pin:
Thanks again,
Dan
Here's a tweaked layout with the 0Vsense pin:
An externally hosted image should be here but it was not working when we last tested it.
Thanks again,
Dan
Looks good but here are a couple of thoughts from me:
C6 is probably to no use at all becasue of the long traces and if you put it nearer the output you would probably get oscillations. I solved this by inserting a small resistor between output and pin 7 of the opamp. By this achieving good decoupling of the opamp and isolate a bit towards the output.
You may move T3 a bit and there draw the trace from the out to the LM317. You need only a small trace like 20-30 mil. By this you will remove a rather long loop. It may not be important.
You can also make some small adjustments in order to get a more compact solution.
T1 up a bit
D1 up and left
T2 left
IC1 left
Out connection up, left
C3 left
and the rest of the package a bit left.
Strive to have a fatter ground trace.
Maybe 100 nF at the input would be appropiate.
I wonder also have much good D3, D4 along with modern opams.
15 V out than a real short circuit.
Pin 7 0 V
Pin 3 +7V
Pin 2 -7 V
Is this enough to damage an opamp?
With diodes:
Pin 7 0 V
Pin 3 around +-1 V
Pin 2 around +- 1 V
and not more than 0.7 volts between the inputs.
If you happen to have more gain than 2 or less then you will get several volts above or below pin 7 at the inputs.
So in real life under normal conditions, how much good do those diodes. They do no harm, I'm sure. Maybe add some capacitance but compared to 500 ohms impedance it's not so much.
C6 is probably to no use at all becasue of the long traces and if you put it nearer the output you would probably get oscillations. I solved this by inserting a small resistor between output and pin 7 of the opamp. By this achieving good decoupling of the opamp and isolate a bit towards the output.
You may move T3 a bit and there draw the trace from the out to the LM317. You need only a small trace like 20-30 mil. By this you will remove a rather long loop. It may not be important.
You can also make some small adjustments in order to get a more compact solution.
T1 up a bit
D1 up and left
T2 left
IC1 left
Out connection up, left
C3 left
and the rest of the package a bit left.
Strive to have a fatter ground trace.
Maybe 100 nF at the input would be appropiate.
I wonder also have much good D3, D4 along with modern opams.
15 V out than a real short circuit.
Pin 7 0 V
Pin 3 +7V
Pin 2 -7 V
Is this enough to damage an opamp?
With diodes:
Pin 7 0 V
Pin 3 around +-1 V
Pin 2 around +- 1 V
and not more than 0.7 volts between the inputs.
If you happen to have more gain than 2 or less then you will get several volts above or below pin 7 at the inputs.
So in real life under normal conditions, how much good do those diodes. They do no harm, I'm sure. Maybe add some capacitance but compared to 500 ohms impedance it's not so much.
janneman said:
Maybe he should read my
Sorry, Jan. I think you are right. It's been such a long time since reading them that I forgot about what belonged to whom.
What I do remember about your article is the question on what chip to use, considering the advantages vs the disadvantages of using the AD797, which proved quite unstable. Time seemed to have confirmed that "treacherous ground".
You also mentioned how tricky it was doing a good design for this project, if I am not wrong.
BTW: very interesting your comments on another thread on whether a ground plane was essential for this design.
Carlos
carlmart said:
I think the main reason for chosing AD797 was the low noise(right? anything else?) . Many years have passed now and there are lot's a good alternatives.
I think performance between OPA134 and THS4601 is the range now and there are many types, altough AD825 seems to be pretty ideal.
Thanks again guys...
Peranders, I've got the layout more compact from your suggestions just a cpl points I need to be clear on...
So with C6 left in (or even if not stuffed) still add a resistor from the opamp V+ to output? I guess I could add this 'offboard' on top of the opamp...but would the AD need a heatsink - from the little I understand I've only seen this done when biasing opamps into classa...
WRT decoupling the opamp - just a 100n on the pin7 to gnd ye?
Do you mean joining the the collecter of T3 and output of the LM317 at the back of the board? So end the signal at the LED ye?
Well for the pos reg I'll be needing 30V so will be going with the AD825 - I guess that means these diodes can go then?
Cheers
Peranders, I've got the layout more compact from your suggestions
just a cpl points I need to be clear on...So with C6 left in (or even if not stuffed) still add a resistor from the opamp V+ to output? I guess I could add this 'offboard' on top of the opamp...but would the AD need a heatsink - from the little I understand I've only seen this done when biasing opamps into classa...
WRT decoupling the opamp - just a 100n on the pin7 to gnd ye?
Do you mean joining the the collecter of T3 and output of the LM317 at the back of the board? So end the signal at the LED ye?
Well for the pos reg I'll be needing 30V so will be going with the AD825 - I guess that means these diodes can go then?
Cheers
daArry said:Thanks again guys...
Peranders, I've got the layout more compact from your suggestions
So with C6 left in (or even if not stuffed) still add a resistor from the opamp V+ to output? I guess I could add this 'offboard' on top of the opamp...but would the AD need a heatsink - from the little I understand I've only seen this done when biasing opamps into classa...
WRT decoupling the opamp - just a 100n on the pin7 to gnd ye?
Do you mean joining the the collecter of T3 and output of the LM317 at the back of the board? So end the signal at the LED ye?
Well for the pos reg I'll be needing 30V so will be going with the AD825 - I guess that means these diodes can go then?
Cheers
The thing to do is check the data sheet and see what the abs max input voltages are AND the abs max differential input voltages. The last data will probably be the limiting factor. From there you can decide if you need the diodes.
Jan Didden
here's the latest layout:
Thanks again
An externally hosted image should be here but it was not working when we last tested it.
Thanks again
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.