I missed the timing window for paying for the Salas BIB group buy because I became ill so I need recommendations for an effective relatively compact PSU kit or at least PCBs with documentation capable of outputting +-30V at 600mA which I'll be using two of to power two SS Dynahi boards. What do you guys recommend, apart from the S22 which is too big?
EDIT: Here are the detailed specifications I want:
1. I'll be using a separate transformer so no need to worry about that. I just need the rectification and AC/DC converter section.
2. I want there to be either user-adjustable voltage with a pot or documentation which helps me chose the voltage through part selection without tooooo much hassle.
3. No positive/negative tracking required. Separately adjustable rails are fine. Don't really care how this part works.
4. Must be capable of an output of +-30V and be capable of at least 300mA of output current continuous per rail, preferably 500mA to allow for overhead.
5. Separate PCBs for the positive and negative sides are fine. The total volume of the combined positive/negative rails must however be smaller than the AMB S22 (6' by 3.5' by 1.5'.)
6. I don't care about what devices are used (BJTs, ICs, JFETs, MOSFETs etc are all fine) as long as it performs well.
7. Must perform at least as good as classic LM317/337-based PSUs in terms of output noise and ripple. Ideal performance roughly on par with the S22, Salas BIB, Goldenreference Low Voltage, etc.
8. Must be readily available in at least bare PCB form with adequate documentation.
9. I will however consider ones which have good, easy to follow documentation which require me to use e.g. protoboard to create the PSU. I'm an idiot, so treat me as such.
10. Must cost at most the price of a Salas BIB to build.
So far, the ones I've found so far are the Salas BIB (which I unfortunately missed the group buy for), the S22 (it's huge though so ideally I want something smaller) and the goldenreference low voltage (GRLV) by Kevin Gilmore (very limited availability just like the Salas BIB unless I do up my own boards and/or start a group buy).
EDIT: Here are the detailed specifications I want:
1. I'll be using a separate transformer so no need to worry about that. I just need the rectification and AC/DC converter section.
2. I want there to be either user-adjustable voltage with a pot or documentation which helps me chose the voltage through part selection without tooooo much hassle.
3. No positive/negative tracking required. Separately adjustable rails are fine. Don't really care how this part works.
4. Must be capable of an output of +-30V and be capable of at least 300mA of output current continuous per rail, preferably 500mA to allow for overhead.
5. Separate PCBs for the positive and negative sides are fine. The total volume of the combined positive/negative rails must however be smaller than the AMB S22 (6' by 3.5' by 1.5'.)
6. I don't care about what devices are used (BJTs, ICs, JFETs, MOSFETs etc are all fine) as long as it performs well.
7. Must perform at least as good as classic LM317/337-based PSUs in terms of output noise and ripple. Ideal performance roughly on par with the S22, Salas BIB, Goldenreference Low Voltage, etc.
8. Must be readily available in at least bare PCB form with adequate documentation.
9. I will however consider ones which have good, easy to follow documentation which require me to use e.g. protoboard to create the PSU. I'm an idiot, so treat me as such.
10. Must cost at most the price of a Salas BIB to build.
So far, the ones I've found so far are the Salas BIB (which I unfortunately missed the group buy for), the S22 (it's huge though so ideally I want something smaller) and the goldenreference low voltage (GRLV) by Kevin Gilmore (very limited availability just like the Salas BIB unless I do up my own boards and/or start a group buy).
Last edited:
I guess a lot depends upon whether you intend to run your final Dynahi amplifier at an idling power of 35 watts per channel. If so then the AMB Dynahi PSU schematic is already debugged, all you have to do is lay it out on a PCB. However it uses ICs; if you prefer a PSU with only discrete devices and no ICs, you may want to consider scaling up Walt Jung's original "Pooge 5.51" voltage regulator designs shown here
How about two Maida style floating regulators. One with "negative" terminal connected to ground so that the "positive" terminal delivers positive regulated output voltage. The other with "positive" terminal connected to ground so that the "negative" terminal delivers negative regulated output.
For 30V DC output (either + or -), use an AC source of approx 40VAC to 50VAC.
For 30V DC output (either + or -), use an AC source of approx 40VAC to 50VAC.
You will have to create a list of required specifications that a kit must meet or exceed, else you cannot / will not buy it. Whether you share this list with the rest of the world is your choice, of course.
Nobody but you knows what specifications need to go on this list; what's important to YOU might not be important to person X, and vice versa.
If I were to create such a list of specifications, for hypothetical person X, it might resemble something like:
~
Nobody but you knows what specifications need to go on this list; what's important to YOU might not be important to person X, and vice versa.
If I were to create such a list of specifications, for hypothetical person X, it might resemble something like:
1. PSU must be dual voltage: Mains input can be either (230V 50Hz) or (115V 60Hz) with a simple jumper option or switch in the primary circuit
2. Output voltage must be somewhere in the range (28V <= Vout <= 33V) for both the positive supply and the negative supply
3. Output voltage must be user adjustable (with a potentiometer on the PCB)
4. It's OK with me if VOUT+ adjustment is separate from the VOUT- adjustment; I do not insist upon "tracking" supplies where one potentiometer controls both VOUT+ and VOUT- simultaneously
5. Output current must be more than 1.2 amperes, continuous, for both VOUT+ and VOUT-. The transformer and the regulator PCB must not overheat when supplying 1.2 amperes for 24 continuous hours.
6. No MOSFETs! I insist that I will not buy any PSU which uses MOSFETs anywhere in the circuit design. Only ICs, JFETs, and BJTs please!
7. Integrated circuits are OK with me. I do not insist upon a fully discrete design.
8. Output ripple must be less than 10mV at 1.2 ampere load
9. Output noise must be less than 50 microvolts in the audio band
Naturally, the specifications that are important to YOU, might be different. But I think you need to decide what you want, before deciding which kit product(s) meet those requirements. 2. Output voltage must be somewhere in the range (28V <= Vout <= 33V) for both the positive supply and the negative supply
3. Output voltage must be user adjustable (with a potentiometer on the PCB)
4. It's OK with me if VOUT+ adjustment is separate from the VOUT- adjustment; I do not insist upon "tracking" supplies where one potentiometer controls both VOUT+ and VOUT- simultaneously
5. Output current must be more than 1.2 amperes, continuous, for both VOUT+ and VOUT-. The transformer and the regulator PCB must not overheat when supplying 1.2 amperes for 24 continuous hours.
6. No MOSFETs! I insist that I will not buy any PSU which uses MOSFETs anywhere in the circuit design. Only ICs, JFETs, and BJTs please!
7. Integrated circuits are OK with me. I do not insist upon a fully discrete design.
8. Output ripple must be less than 10mV at 1.2 ampere load
9. Output noise must be less than 50 microvolts in the audio band
~
Here's my list. I already know what I want, I just forgot to specify exactly what.
1. I'll be using a separate transformer so no need to worry about that. I just need the rectification and AC/DC converter section.
2. I want there to be either user-adjustable voltage with a pot or documentation which helps me chose the voltage through part selection without tooooo much hassle.
3. No positive/negative tracking required. Separately adjustable rails are fine. Don't really care how this part works.
4. Must be capable of an output of +-30V and be capable of at least 300mA of output current continuous per rail, preferably 500mA to allow for overhead.
5. Separate PCBs for the positive and negative sides are fine. The total volume of the combined positive/negative rails must however be smaller than the AMB S22 (6' by 3.5' by 1.5'.)
6. I don't care about what devices are used (BJTs, ICs, JFETs, MOSFETs etc are all fine) as long as it performs well.
7. Must perform better than classic LM317/337-based PSUs in terms of output noise and ripple. Ideal performance roughly on par with the PSUs I've already found.
8. Must be readily available in at least bare PCB form with adequate documentation.
9. I will however consider ones which have good, easy to follow documentation which require me to use e.g. protoboard to create the PSU. I'm an idiot, so treat me as such.
10. Must cost at most the price of a Salas BIB to build, including the cost of all PCBs.
So far, the ones I've found so far are the Salas BIB (which I unfortunately missed the group buy for), the S22 (it's huge though so ideally I want something smaller) and the goldenreference low voltage (GRLV) by Kevin Gilmore (very limited availability just like the Salas BIB unless I do up my own boards and/or start a group buy).
1. I'll be using a separate transformer so no need to worry about that. I just need the rectification and AC/DC converter section.
2. I want there to be either user-adjustable voltage with a pot or documentation which helps me chose the voltage through part selection without tooooo much hassle.
3. No positive/negative tracking required. Separately adjustable rails are fine. Don't really care how this part works.
4. Must be capable of an output of +-30V and be capable of at least 300mA of output current continuous per rail, preferably 500mA to allow for overhead.
5. Separate PCBs for the positive and negative sides are fine. The total volume of the combined positive/negative rails must however be smaller than the AMB S22 (6' by 3.5' by 1.5'.)
6. I don't care about what devices are used (BJTs, ICs, JFETs, MOSFETs etc are all fine) as long as it performs well.
7. Must perform better than classic LM317/337-based PSUs in terms of output noise and ripple. Ideal performance roughly on par with the PSUs I've already found.
8. Must be readily available in at least bare PCB form with adequate documentation.
9. I will however consider ones which have good, easy to follow documentation which require me to use e.g. protoboard to create the PSU. I'm an idiot, so treat me as such.
10. Must cost at most the price of a Salas BIB to build, including the cost of all PCBs.
So far, the ones I've found so far are the Salas BIB (which I unfortunately missed the group buy for), the S22 (it's huge though so ideally I want something smaller) and the goldenreference low voltage (GRLV) by Kevin Gilmore (very limited availability just like the Salas BIB unless I do up my own boards and/or start a group buy).
Last edited:
Yup, then I think one option for you will be Pete Millet's Maida PSU board. You'd need two of them (as discussed) for a dual rail supply. He has an ebay store I think, but also, here's a link to his web site that describes it: (no affiliation) High Voltage Regulator
To get 30V output, you'd need to set R5, R4, and RV1 to the values you need, and as discussed use something like a 40V AC source. Everything else would be exactly as shown on the schematic.
The relationship that describes the output voltage is: Vout = 1.25(1 + RX/R5), where RX = R4+RV1.
If you made R5 100 ohms, R4 + RX would need to be 2.3K, so you could go with a 1K variable pot + 1.3K fixed resistor. Tweak the size of those values to get you a little more or little less variation, as long as the sum of them is 2.3K. You could even use fixed resistors for both and be locked right near 30V output (given tolerance of the resistors used).
If you use a 40V AC source, the mosfet would need to drop about 20V @ 300 mA, which will generate approx 6 watts of heat, so mounting the mosfet on a moderately sized heat sink will be required. But I'm guessing you would have had the same heatsinking issues anyway even if you used the Salas regulator.
By the way, there was another thread presented here a while back called the "21st Century Maida regulator" or something akin to that, which is in essence a Maida regulator but made with a newer regulator IC that has quite a bit better noise performance than the 317, but I'm not sure what max sustained output current it can support. I think if you performed a search you could easily find that thread on here somewhere.
To get 30V output, you'd need to set R5, R4, and RV1 to the values you need, and as discussed use something like a 40V AC source. Everything else would be exactly as shown on the schematic.
The relationship that describes the output voltage is: Vout = 1.25(1 + RX/R5), where RX = R4+RV1.
If you made R5 100 ohms, R4 + RX would need to be 2.3K, so you could go with a 1K variable pot + 1.3K fixed resistor. Tweak the size of those values to get you a little more or little less variation, as long as the sum of them is 2.3K. You could even use fixed resistors for both and be locked right near 30V output (given tolerance of the resistors used).
If you use a 40V AC source, the mosfet would need to drop about 20V @ 300 mA, which will generate approx 6 watts of heat, so mounting the mosfet on a moderately sized heat sink will be required. But I'm guessing you would have had the same heatsinking issues anyway even if you used the Salas regulator.
By the way, there was another thread presented here a while back called the "21st Century Maida regulator" or something akin to that, which is in essence a Maida regulator but made with a newer regulator IC that has quite a bit better noise performance than the 317, but I'm not sure what max sustained output current it can support. I think if you performed a search you could easily find that thread on here somewhere.
I would recommend one of Kevin Gilmore’s GRLVs. These will do +/-30V at 1A or better, and are smaller than an S22. Very low noise (Kevin had to take one into a screen room at Northwestern to measure it.)
goldenreference low voltage power supply - Do It Yourself - www.Head-Case.org
I do have some of the original Dynahi PSU pcbs if you want to go that route (the ones using OPA548s).
goldenreference low voltage power supply - Do It Yourself - www.Head-Case.org
I do have some of the original Dynahi PSU pcbs if you want to go that route (the ones using OPA548s).
Last edited:
That was what I was wondering. I would've gone with the GRLV right off the bat but there doesn't seem to be any boards for sale or any sort of group buy ongoing.
The Gerbers are readily available... you could get some boards run at someplace like Seeedstudio for pretty cheaply. I'd use 2 oz copper, but otherwise their standard options. If you went with the 3, 4 or 5 board gerbers, these are at or under 100mm x 100mm, so a bit cheaper.
Link to gerbers:
staxpcbs - Google Drive
The plus and minus versions are single polarity, so those aren't the ones you want.
Link to gerbers:
staxpcbs - Google Drive
The plus and minus versions are single polarity, so those aren't the ones you want.
Last edited:
Why not just get a transformer with CT and use a bipolar regulator like this one. The part description is misleading, its actually a dual output regulator. Have a look at the spec for it.
Adjustable Voltage Regulator Power Supply Board +/- 1.25V to 37V DC Output LM317/LM337 Design
If you need cleaner add a simple PI filter of some sort.
Adjustable Voltage Regulator Power Supply Board +/- 1.25V to 37V DC Output LM317/LM337 Design
If you need cleaner add a simple PI filter of some sort.
Last edited:
I have considered it but I can't tell the difference between a well laid-out design and a not so well laid-out design. I'd ideally want one which is well laid-out, so.
The designer that made that regulator knew what they were doing. It's a bargain.
Generous track thickness in the layout. Uses 2 parallel capacitors (bulk + decouple ) to get wider usable frequency range. Uses reverse bias clamping diodes across regulators and outputs. Heat sinks on the regulators.
The LM series regulators may not sufficiently reject some types of HF line noise (>10KHz) which is usually not a problem. That's why I mentioned a PI filter if required.
Generous track thickness in the layout. Uses 2 parallel capacitors (bulk + decouple ) to get wider usable frequency range. Uses reverse bias clamping diodes across regulators and outputs. Heat sinks on the regulators.
The LM series regulators may not sufficiently reject some types of HF line noise (>10KHz) which is usually not a problem. That's why I mentioned a PI filter if required.
if I'm going for one of those LM317 SUs, I do want something that's available in the UK because it's such an ubiquitous design. Do you know of any similarly laid out designs available over there?