Thanks AndrewT!
I used 51k since 47k was in the original schematic. Do you think it can cause any problems (drops due to input bias current) or you suggest that just a small one would do the job?
C3 is planned to be "attached" to the "input pins" of the regulator, speaking of layout - I know that I have not indicated that too well in this schematic, it seems like it is attached to the second electrolytic! Is this why you say it is in the wrong place? It is the typical 100n cap that we fit right at the input of regulators to bump transients. C5 will be attached right at the NE5534 pins.
I used 51k since 47k was in the original schematic. Do you think it can cause any problems (drops due to input bias current) or you suggest that just a small one would do the job?
C3 is planned to be "attached" to the "input pins" of the regulator, speaking of layout - I know that I have not indicated that too well in this schematic, it seems like it is attached to the second electrolytic! Is this why you say it is in the wrong place? It is the typical 100n cap that we fit right at the input of regulators to bump transients. C5 will be attached right at the NE5534 pins.
Last edited:
I think you may want to install a startup circuit since your reference voltage is derived from your output voltage. Samuel Groner's power supply schematic shows one way to do it.
I also think it's prudent to assume the opamp input bias current will be 3X the datasheet max. For a 5534 that's 3 x 1500nA = 4.5 microamps. Use this to calculate the voltage error introduced by resistor R8.
I also think it's prudent to assume the opamp input bias current will be 3X the datasheet max. For a 5534 that's 3 x 1500nA = 4.5 microamps. Use this to calculate the voltage error introduced by resistor R8.
If it is a non-changing error, I can live with that! That's why the trimmer is used, I don't care fot the exact value of the reference at the positive input, as long as it stays constant. 🙂
Original Sulzer regulator schematic does not use any startup circuit - I also read a Jung article where he implemented a Sulzer regulator with no startup too. Do you think any problems could rise?
Original Sulzer regulator schematic does not use any startup circuit - I also read a Jung article where he implemented a Sulzer regulator with no startup too. Do you think any problems could rise?
R8 can be too large only if you have a leaky cap at the end; the leakage current might cause a voltage drop. Otherwise, the larger the better.
The startup circuit is typically necessary if the opamp supply comes from the output. In my experience, this Sulzer type circuit *should* start up fine.
Jan
About the resistance change due to heating - take a 5W resistor of 400 PPM/C, and one of 20 PPM/C. How more stable will the second one be? I mean, will one notice any real difference in value for up to, say, 70C ambient?
And is it really crucial for an RC filter after the reservoir? My guess is that it doesn't matter that much.
And is it really crucial for an RC filter after the reservoir? My guess is that it doesn't matter that much.
I updated the schematic a bit. Here it is:
I have also made a first layout for it. It is narrow since I want to put 12 in series. The layout is as viewed from above (components side), single-sided pcb. Rf, Rg and 3.3k are mounted vertically. To start, I attach only the regulator side (not the fuse, bridge, RC filter part). There it is:
I have made basic connections only - not fatten up tracks yet. C3 and C5 (if used) shall be mounted on the copper side. Any opinions? 🙂
I have also made a first layout for it. It is narrow since I want to put 12 in series. The layout is as viewed from above (components side), single-sided pcb. Rf, Rg and 3.3k are mounted vertically. To start, I attach only the regulator side (not the fuse, bridge, RC filter part). There it is:
I have made basic connections only - not fatten up tracks yet. C3 and C5 (if used) shall be mounted on the copper side. Any opinions? 🙂
C4 must be at least 50V.
The other tantalum must be at least twice the peak voltage they will be subjected to.
This doubling of voltage seems to be the necessary method of ensuring reliable life.
The other tantalum must be at least twice the peak voltage they will be subjected to.
This doubling of voltage seems to be the necessary method of ensuring reliable life.
Thanks for your comment.
Then I might consider using good electrolytic caps. 🙁 Having to buy tantalums over 35V at 47uF for 12 supplies is completely out of my budget.
How about C8? For a 9V supply, would 25V be adequate? For a 16V, a 35V?
Also, do you have any remarks on my layout?
Audiostrat,
R8 should equal Rf||Rg.
If R8 value is decreased, C7 value should be increased, to keep LPF frequency low.
Input current noise, 0.6pA over 51K.....your current noise will dominate over voltage noise. MT-047 Tutorial from Analog Devices will give you more info.
And 0.1€ for 10 - 47R resistor between opamp and Base of pass transistor is very well spent money.
R8 should equal Rf||Rg.
If R8 value is decreased, C7 value should be increased, to keep LPF frequency low.
Input current noise, 0.6pA over 51K.....your current noise will dominate over voltage noise. MT-047 Tutorial from Analog Devices will give you more info.
And 0.1€ for 10 - 47R resistor between opamp and Base of pass transistor is very well spent money.
To keep costs down, C4, C6, C8 can be electrolytic capacitor, no need for tantalum.
C7 can be lytic too, but choose low leakage type.
C7 can be lytic too, but choose low leakage type.
Many thanks Stormsonic! Is there an absolute value for the current limiting resistor to base?
Also, what you say about R8 is very interesting and as I see is used in Walt Jung's Super regulator. Why is that beneficial?
Did you also find any flaws in my proposed layout? 🙂
EDIT: Thanks for the link too.
Also, what you say about R8 is very interesting and as I see is used in Walt Jung's Super regulator. Why is that beneficial?
Did you also find any flaws in my proposed layout? 🙂
EDIT: Thanks for the link too.
Last edited:
I had a look at Linear Audio graphs regarding the comparison of output noise for positive regulators.
But I see that the Sulzer (I assume the original topology) has better noise specs than Jung Super regulator.
So, how much would really matter not using the 47k resistor the original circuit used for filtering of the reference? The circuit seems to have already fantastic specs regarding noise. Don't we gain reliability usinf the 4.7uF tantalum + 47k instead of using a 100-120uF electolytic?
On the other hand, could using high capacitance for the feedback section and this rule (R8 equals Rf||Rg) improve line rejection or anything else? It just seems to me that discarding the original 47k is not a matter of noise at the end of the day. Is that true?
But I see that the Sulzer (I assume the original topology) has better noise specs than Jung Super regulator.
So, how much would really matter not using the 47k resistor the original circuit used for filtering of the reference? The circuit seems to have already fantastic specs regarding noise. Don't we gain reliability usinf the 4.7uF tantalum + 47k instead of using a 100-120uF electolytic?
On the other hand, could using high capacitance for the feedback section and this rule (R8 equals Rf||Rg) improve line rejection or anything else? It just seems to me that discarding the original 47k is not a matter of noise at the end of the day. Is that true?
I think that it would be much better to keep the traces separate, going into and out of the large electrolytic cap's pads. i.e. Don't have a single "stub" from each of the pads, that then divides into upstream and downstream traces, and is shared by both the upstream and downstream currents. The noisier cap input current will induce voltages across the stub trace, which allows them to get into the downstream circuitry.
Gootee, thanks for the reply, but I don't think I understand what you mean. 🙁
In my schematic I have spiltted the series resistor into both supply rails, to isolate the noisy currents from the first reservoir. I don't really know if this is what you are talking about, or something that I have done on my layout wrong.
Seems like my English betrayed me! 🙂
In my schematic I have spiltted the series resistor into both supply rails, to isolate the noisy currents from the first reservoir. I don't really know if this is what you are talking about, or something that I have done on my layout wrong.
Seems like my English betrayed me! 🙂
There must be a trace carrying charging current into the capacitor.
There must be a separate trace taking DC out of the capacitor.
i.e. an in and an out trace for every capacitor.
This is the basis of the very expensive 4 wire capacitors. They separate the In and Out wires.
There must be a separate trace taking DC out of the capacitor.
i.e. an in and an out trace for every capacitor.
This is the basis of the very expensive 4 wire capacitors. They separate the In and Out wires.
So, for each capacitor pad, the incoming trace, as well as the outgoing, should connect directly to the pad. There should be no common trace for this two traces. Which I have not implemented in my layout yet (post 146).
Or are you suggesting something else? 🙂
Or are you suggesting something else? 🙂
Thank you for the nice observation.
After the reservoirs? Should I follow this logic again, or should I fill as much ground as possible, actually making a big ground "pool" where everything will connect on the regulator side?
Of course, following the original layout connections.
After the reservoirs? Should I follow this logic again, or should I fill as much ground as possible, actually making a big ground "pool" where everything will connect on the regulator side?
Of course, following the original layout connections.
- Status
- Not open for further replies.