Happy boxing day as well. I am now listening to Bill Evens "Bill Evens Alone", 9pm Sydney time...
Tomorrow I will build the latest version of the regulator with my newly acquired larger heatsinks so that the current can go up to 350mA. I will also use two high frequency chokes to make the raw supply CLC.
I have now learnt that the regulator should be placed as close to the circuit as possible, possibly even in the case of remote sensing. With my large circuit boards, the only way to do this is to have multiple regulators. I am planning to have 3 if they fit otherwise two.
I think in theory I should be able to use a single CCS feeding multiple shunts, but in that case the cable running from the CCS to the shunts have to be very long. Would that be a problem?
Regards,
Bill
Tomorrow I will build the latest version of the regulator with my newly acquired larger heatsinks so that the current can go up to 350mA. I will also use two high frequency chokes to make the raw supply CLC.
I have now learnt that the regulator should be placed as close to the circuit as possible, possibly even in the case of remote sensing. With my large circuit boards, the only way to do this is to have multiple regulators. I am planning to have 3 if they fit otherwise two.
I think in theory I should be able to use a single CCS feeding multiple shunts, but in that case the cable running from the CCS to the shunts have to be very long. Would that be a problem?
Regards,
Bill
Last edited:
Hey Bill, you're back for real!?!
Regarding one CCS and multiple shunts with long wires, this is something I haven't tried. How long is "very long" in your case? Meters? One thing to keep in mind is that a long wire that's not shielded makes a very nice antenna!
Everybody, sorry about the lack of updates, I was foolish to think that I'd have a lot of time this past week, but all my time was sucked up by the Christmas preparations, shopping, and all that goes with it. One of those non-negotiable fact of life, family first.
Anyway, I'm having trouble adding anything to the wiki, it seems stuck in some weird state of confirmation of some edit. It's totally stuck. I don't know how to fix that but it seems to have happened earlier to other GBs and then the list was done in the thread. I'd like to keep the thread free of hundreds of messages about "add me too please" so I'll setup very soon a website over which I have control to fix things where people can add themselves to show their interest. As I said before, it would not be an engagement to buy anything.
Regarding one CCS and multiple shunts with long wires, this is something I haven't tried. How long is "very long" in your case? Meters? One thing to keep in mind is that a long wire that's not shielded makes a very nice antenna!
Everybody, sorry about the lack of updates, I was foolish to think that I'd have a lot of time this past week, but all my time was sucked up by the Christmas preparations, shopping, and all that goes with it. One of those non-negotiable fact of life, family first.
Anyway, I'm having trouble adding anything to the wiki, it seems stuck in some weird state of confirmation of some edit. It's totally stuck. I don't know how to fix that but it seems to have happened earlier to other GBs and then the list was done in the thread. I'd like to keep the thread free of hundreds of messages about "add me too please" so I'll setup very soon a website over which I have control to fix things where people can add themselves to show their interest. As I said before, it would not be an engagement to buy anything.
Merry Christmas, Happy Hanukkah, Happy Holidays, whatever you celebrate, may you have a great and joyous time!
Happy holidays!!!
I will think of single CCS with multiple shunts later when the final circuit board is designed. I wasn't worried about noise but inductance. But now first thing first. I will build the 5c/5d and use it for final driver measurements.
I have noticed that R7 is 750R with 5c and 330R with 5d. You said before that the absolute value is not important but there is an optimal value Idss for stable temporature. I guess it is the same for the R10 = 1.2k.
My R2 will be a trimpot for the adjustment of CCS. For adjustment of the shunt current if I fix R10 = 1.2k and use two trimports for R11, R9 it is rather tricky to get R11 and R9 the same value. So I guess I will fix R11 and R9 instead and make R10 a trimpot as long as the value falls somewhere between 500R to 1.5k. What do you think?
I think trimports here for permanent use should be fine as the current going through them would be very small. It is tricky to get the exact value resistors unless I put 3 or 4 in parallel / series that may even increase noise pickup. I hate to solder multiple resistors in some tiny space.
I have noticed that R7 is 750R with 5c and 330R with 5d. You said before that the absolute value is not important but there is an optimal value Idss for stable temporature. I guess it is the same for the R10 = 1.2k.
My R2 will be a trimpot for the adjustment of CCS. For adjustment of the shunt current if I fix R10 = 1.2k and use two trimports for R11, R9 it is rather tricky to get R11 and R9 the same value. So I guess I will fix R11 and R9 instead and make R10 a trimpot as long as the value falls somewhere between 500R to 1.5k. What do you think?
I think trimports here for permanent use should be fine as the current going through them would be very small. It is tricky to get the exact value resistors unless I put 3 or 4 in parallel / series that may even increase noise pickup. I hate to solder multiple resistors in some tiny space.
I'd worry about it too. I don't know how long you can get away with, it's not something I was going to build. You have quite a different setup.
I will sync the negative and positive boards and schematics to 5d both, just haven't had the time yet. Without measuring the jfet it's hard to tell which value is best for R7 or R10. 330R, 470R, all work fine. I'd suggest not to obsess about these values, I've said it before.
R9 and R11 don't have to be exactly the same. IMHO these are not the parts of the circuit. If I were you I'd make R10 fix and only one of R9 and R11 a trimpot, the other fix too.
The voltage drifts a bit with temperature, which voids your effort to get an exact value resistor. I'd use a trimpot to get the value wanted, then replace it with an approximately equal value fixed resistor. In my experiments I have used a trimpot ONLY for R2, which sets the current limit. Everywhere else I've used fixed values which I then replaced until the voltage out was closest to what I wanted. Still, one good quality trimpot would be ok.
Everybody else, I setup another page where you can add yourselves to show interest in getting pcbs; once I have a total number of boards wanted which I can get a quote from the pcb manufacturer. This is the link.
Here's 5d positive for completeness.
An externally hosted image should be here but it was not working when we last tested it.
Hi ikoflexer,
on your wishes i have added the PCB design text. Also i added the RS and two pots for R2 & R9.
There are still some questions. Because i will make a 2-Layer design i could separate the GND-Line to the Top-Layer (blue line).
There i could make also the GND-Line over the complete PCB. Do you think this is an advantage?
BTW, is there a fixed formula to calculate R2 (CCS current) & R9 (Vout) only if i have Vin and know the needed current output?
Best regards and a happy new year,
Oliver
on your wishes i have added the PCB design text. Also i added the RS and two pots for R2 & R9.
There are still some questions. Because i will make a 2-Layer design i could separate the GND-Line to the Top-Layer (blue line).
There i could make also the GND-Line over the complete PCB. Do you think this is an advantage?
BTW, is there a fixed formula to calculate R2 (CCS current) & R9 (Vout) only if i have Vin and know the needed current output?
Best regards and a happy new year,
Oliver
Attachments
Last edited:
No, there is no fixed value for R2.
If you want MOSFET M2 to start conducting, you need some voltage between Gate and Source, this is called Vgs treshold (look at datasheet). More voltage you apply, higher current will be through M2.
You will have some voltage drop on D1 and some voltage drop on R2. Voltage drop on R2 depends from current through JFET J3 and J4 CCS. Current through J3 and J4 depends on resistor R7. All JFETs are not equal (different Idss) and all MOSFETs are not equal, different Vgs. Too much variables to calculate fixed value for R2. Better to use trimmer, then replace with fixed value.
Same situation with R9. Voltage drop on R9 depends on current through R9. Current through R9 depends on R11 value, base currents of Q2, Q3 and J2. Current through J2 depends on JFET used and R10 resistor
If you want MOSFET M2 to start conducting, you need some voltage between Gate and Source, this is called Vgs treshold (look at datasheet). More voltage you apply, higher current will be through M2.
You will have some voltage drop on D1 and some voltage drop on R2. Voltage drop on R2 depends from current through JFET J3 and J4 CCS. Current through J3 and J4 depends on resistor R7. All JFETs are not equal (different Idss) and all MOSFETs are not equal, different Vgs. Too much variables to calculate fixed value for R2. Better to use trimmer, then replace with fixed value.
Same situation with R9. Voltage drop on R9 depends on current through R9. Current through R9 depends on R11 value, base currents of Q2, Q3 and J2. Current through J2 depends on JFET used and R10 resistor
Hi Ikoflexer,
I have been following this thread with much interest since a while.
Unfortunately my projects progress very very slowly because of lack of time...
I have a, perhaps dumb, question.
Is it possible to put the shunt element between two CCS's (positive and negative)? So the shunt element is between the positive and negative rail?
Doubling the power on the Shunt element of course. I can imagine there will be some drawbacks, but since your the expert and are able to simulate...
Reason:
I want to use your regulator for an active filter (many thanks mister Linkwitz).
Opamps only need positive and negative supply. If possible I want to avoid big currents through my grounds,and it makes PCB layout also a lot easier.
Thanks for your design.
Marcel
I have been following this thread with much interest since a while.
Unfortunately my projects progress very very slowly because of lack of time...
I have a, perhaps dumb, question.
Is it possible to put the shunt element between two CCS's (positive and negative)? So the shunt element is between the positive and negative rail?
Doubling the power on the Shunt element of course. I can imagine there will be some drawbacks, but since your the expert and are able to simulate...
Reason:
I want to use your regulator for an active filter (many thanks mister Linkwitz).
Opamps only need positive and negative supply. If possible I want to avoid big currents through my grounds,and it makes PCB layout also a lot easier.
Thanks for your design.
Marcel
Yes, received email today
I think problem was in different MOSFETs (IRFBC40 & CEB703AL).
Input 12V, shunt current set to 500 mA. Output 5V, load 100 mA (resistor).
Look at 2nd picture, lower left corner. Grid is 5 mV, output ripple 3.6 mV peak-to-peak.
I don't know if you have solved this problem or not.
If you use an electrolytic capacitor at the output of the regulator close to the MOSFET, you won't get any resonance from the regulator. I have done this a few times. This can be tested with a resistor dummy load.
However, I found when connecting to an opamp there was a resonance. This resonance is not caused by the regulator but the wire inductance from the MOSFET to the opamp power inputs, as well as the capacitance at the opamp inputs. You can model this with LTSpice. Even if you take out the regulator, the wire inductance and the input capacitance will still cause resonance.
I connected a 22uF in series with a 1R metal film resistor at the opamp power inputs (shunt to ground) and the resonance was removed. You may try this one.
Hi Marcel,
Sorry I didn't notice your post until a few minutes ago, I was kind of busy in another thread for a couple of days. I'm running out the door right now but I'll come back with an answer later tonight.
Thanks!
I have been very busy these days and have not had the time to write detailed reports. Here is a brief one.
It took me a week to rebuild my regulator to 5d! Much harder than I expected.
I would highly recommend people to buy the PCB and save all the trouble and time, instead of trying your own. I have had no choice because I want the regulators now.
To get around 15V output, R10=1k2, R11=18k, R9 is 20k trimpot with value set above 15k.
For the CCS, R7=330R, R2=1.5k + (1k)trimpot (25 turns). Even so, the trimport is still very sensitive. If you use larger trimpot without the 1.5k resistor, it will be so sensitive that it is extremely difficult to get the CCS value you want.
It took me a week to rebuild my regulator to 5d! Much harder than I expected.
I would highly recommend people to buy the PCB and save all the trouble and time, instead of trying your own. I have had no choice because I want the regulators now.
To get around 15V output, R10=1k2, R11=18k, R9 is 20k trimpot with value set above 15k.
For the CCS, R7=330R, R2=1.5k + (1k)trimpot (25 turns). Even so, the trimport is still very sensitive. If you use larger trimpot without the 1.5k resistor, it will be so sensitive that it is extremely difficult to get the CCS value you want.
The output capacitor should be placed close to the MOSFET.
I have never found any problems using an electrolytic output capacitor (i.e. no resonance). I guess this is by default.
However, I would like to use film output capacitor. With film capacitor I spent quite a bit of time to get it right.
I initially used what I used in 5a (first version) with 8.8uF MKT. It was mounted close to the MOSFET but this time placed between the two heatsinks. No resonance was found.
I thought the heat was no good for capacitors so I moved the 8.8uF MKT out, about 5cm away from the MOSFET. The regulator resonated.
I then tried using 2.2uF MKT with about 15mm legs soldered at the back of the PCB close to the MOSFET. No resonance was found. I was happy.
I then tidied it up and shortened the 15mm legs to 8mm. I was then shocked. The regulator resonated.
I then used a new pair with 22mm legs and there was a tiny resonance.
I then tried the Visual MKP 1.1uF, it resonated.
After all these experiments, I finally soldered the 8.8uF (4 in parallel) MKT at the back of the PCB with 1cm leg length. There was no resonance. This is the final value I chose.
For those who are not as crazy as me (to insist on using film cap), you would use an electrolytic capacitor at the output. The regulator is very stable with it. I have never found any problem.
If you use film capacitor, it may be marginally stable.
I have never found any problems using an electrolytic output capacitor (i.e. no resonance). I guess this is by default.
However, I would like to use film output capacitor. With film capacitor I spent quite a bit of time to get it right.
I initially used what I used in 5a (first version) with 8.8uF MKT. It was mounted close to the MOSFET but this time placed between the two heatsinks. No resonance was found.
I thought the heat was no good for capacitors so I moved the 8.8uF MKT out, about 5cm away from the MOSFET. The regulator resonated.
I then tried using 2.2uF MKT with about 15mm legs soldered at the back of the PCB close to the MOSFET. No resonance was found. I was happy.
I then tidied it up and shortened the 15mm legs to 8mm. I was then shocked. The regulator resonated.
I then used a new pair with 22mm legs and there was a tiny resonance.
I then tried the Visual MKP 1.1uF, it resonated.
After all these experiments, I finally soldered the 8.8uF (4 in parallel) MKT at the back of the PCB with 1cm leg length. There was no resonance. This is the final value I chose.
For those who are not as crazy as me (to insist on using film cap), you would use an electrolytic capacitor at the output. The regulator is very stable with it. I have never found any problem.
If you use film capacitor, it may be marginally stable.
Last edited:
Of course, my implementation is different from Ikoflexer's. With his PCB, it most likely works a lot better than my implementation using veroboard. I used 22awg (0.71mm diameter) wires to wire up the PCB except for the power tracks. I think I will rebuild it with 18-20awg wires and I now believe it will make a big difference. I will use 16awg wires for the power tracks.
In my previous version, the raw supply used 3 x 3,300uF caps in parallel.
In this version, I inserted a 470uH high frequency choke in series with a 1R resistor after the first 3,300uF.
This sounds BETTER!
I look forward to getting a C-Multiplier from Ikoflexer one day.
The regulator now sounds marginally better than 5a.
It sounds very good!
In this version, I inserted a 470uH high frequency choke in series with a 1R resistor after the first 3,300uF.
This sounds BETTER!
I look forward to getting a C-Multiplier from Ikoflexer one day.
The regulator now sounds marginally better than 5a.
It sounds very good!
Marcel, OK, here's what I think about your proposal. In principle you can have just one shunt element connected between the positive and negative rails. But, then there must be one reference+amp to drive this shunt element, and this part of the circuit also has to be connected between the two rails. The results is that the voltage between the positive rail and the negative rail is kept at some constant value, but not the voltage of each rail wrt to the ground. In fact I would expect it to be quite unstable.
In my opinion this is not how you should address grounding problems. Instead, opt for a properly designed pcb and then properly hooked up to the load, which in your case are opamps.
Look at what hifinutnut has done. His load is a bunch of opamps, right Bill?
I think it's the same with stormsonic, he's powering a DAC.
If it's done properly, the opamps will be happy. You can always post your grounding scheme here and we can opinionate about it. You'll be surprised how much help you can get from the community. I certainly try to help as much as I can. You don't have to buy a pcb to get help, we just help. Salas has helped all of us so much, you he deserves a statue. I'm not kidding.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.