Need help in regulating +/-36V regulated supplies

[yungman]

I got tripped up on the voltage regulation design. I just layout and got the pcb back, I am not happy with the result. The application is rather low current of less than 40mA, so I tried to be lazy, use unregulated supply and use a lot of capacitor to filter the ripple out. The problem is

1) The output voltage of the power transformer I bought varies from 50VDC when no load to 44V loading with 30mA.https://www.mouser.com/ProductDetail/530-LP-56-100, I bought 12 of them already so I am stuck with this one.

2) I try to use zener diodes to drop the voltage down to 37V or so. I use 1N5333 3.3V 5W zener. That stupid diode only drop less than 2V at 30mA current!

So between big variation of the output voltage of the transformer and the unpredictable voltage of the zener, I don't feel confident to work of this design anymore. So I am planning to start over again.

I want to use voltage regulator this time. The transformer being at +/-50V at no load, I need both +ve and -ve regulator that can withstand input voltage at least 60V. I have no problem finding LM317HVT that can have +60V input.https://www.digikey.com/product-detail/en/texas-instruments/LM317HVT-NOPB/LM317HVT-NOPB-ND/212662, but I cannot find a negative regulator.

Please advice as I am press for time. Any way to salvage the existing board would be helpful, if not, please advice the easiest way to get the -37V with -50V input regulator. Attached is the schematic of my pcb, the top two circuits are the +/-37V supplies.

Thanks

 

[yungman]

I have some time looking into the LM337, it is supposed to be -40V input max, but from reading the datasheet, there is a way to make it work at 60V input because all 3 terminals are not grounded, all I have to do is to make sure the output and the adj terminal is less than 40V from the input and it should be safe. Here is my modification, please take a look to see whether this will work.

Basically, I put two 33V zener diode to make sure the output and adj will not have over 33V from the input. I pick 33V zener because at low current, it will start to turn on at much lower voltage, I basically assume the zener will conduct at 1/2 the 33V or 16.5V. I can even use 36V or even 39V zener if needed.

Thanks

 

[yungman]

I change the circuit a little bit, too bad I cannot edit the last post anymore.

I use 3 diodes to clamp the adj so I only have to have one zener.

One can never take for granted that the circuit is too simple, never thought I'd be tripped by such a simple power supply circuit. Never thought the zener will drop so little from the spec voltage at lower current. Like a 3.3V zener drops only like 1.5V drawing 30mA! I have to use those 5W zener because I have over 12,000uF of filter caps. Power up current surge can be quite high and I don't think the 1W zener can be reliable long term. But the 5W zener drops too little voltage. More importantly, I don't trust the voltage is repeatable and temperature stable. Or else, I can use say a 6V zener and drop 3V. That's not a good way to design.

Don't laugh at all the zeners on the right part of the circuit. This part is only going to power up the front end section of the power amp, I have a whole different big transformer with 30,000uF filter caps to power up the output section that draws min of 1A. I have to keep the voltage of the front end and output section to within a few volts or else I can burn some circuits, that's where all the zener diodes on the right side comes in, to keep the too power supplies to within 6V. Obviously that goes out the window also as those zener voltages are unpredictable also. It is important for me to isolate the voltages but track to within a few volts.

If anyone have better idea, let me know.

 

[jim hardy]

how about a LM317HV - good for 50 volts?

http://www.ti.com/lit/ds/symlink/lm137hvqml.pdf

ONSEMI seems to have it in the package you need.

 

[yungman]

Hi Jim

Yes, I looked into that already. The LM337HV is only good for -50V. The no load voltage of the transformer is a little over -50V. That's the reason I have the design in post #3 to see whether that works or not. Too bad they don't make it to 60V like the LM317HV.

Thanks

Alan

 

[yungman]

I did the preliminary design attached. Please comment.

 

[jim hardy]

Are your Zeners providing reverse polarity protection for the regulators ?
If so
Make sure they'll conduct both directions - some Zeners won't because they have another junction inside to give better temperature performance

 

[NascentOxygen]

I want to use voltage regulator this time. The transformer being at +/-50V at no load, I need both +ve and -ve regulator that can withstand input voltage at least 60V. I have no problem finding LM317HVT that can have +60V input. https://www.digikey.com/product-detail/en/texas-instruments/LM317HVT-NOPB/LM317HVT-NOPB-ND/212662, but I cannot find a negative regulator.

If the transformer secondaries are independent, can you build a pair of identical floating "positive" DC supplies then connect them so the box presents (+) (0) (–) at its output?

 

[jim hardy]

i don't understand the reason behind the 50 volt DF005M bridge rectifiers BR5 and BR6 connected in parallel with the 150 volt KMB215S ones BR1 and BR2.

What did i miss?

 

[yungman]

i don't understand the reason behind the 50 volt DF005M bridge rectifiers BR5 and BR6 connected in parallel with the 150 volt KMB215S ones BR1 and BR2.

What did i miss?

If you look at the schematic, I have Connector J25 and J27. I want to have the option to just take the AC secondary from the big transformer that power the output section. I want the voltage of the frontend to be a little higher than the output section. If I take the AC from the same transformer, I would use a Schottky bridge rectifier instead of the standard rectifier to gain about 0.4V. I layout so the two bridge rectifier are on top of each other so I can either solder one or the other.

I put in a lot of options on the board that I can power the frontend by a different transformers or by the big power transformers. This is a fail safe so if the I can't use the secondary transformer for whatever reason ( like the zeners are giving me all the trouble), I can just power everything with the main transformer. Most amps don't have separate transformers to power up the frontend and the output section. I literally have 4 power transformers in one amp. Two for the frontend and two for the output sections.

Believe it or not, I can hear the improvement using separate transformers. That's the reason why some of the high end power amps come in Monoblock, you use separate power amp for each left and right channel.

 

[jim hardy]

I see. Either/Or.

Still i'd be afraid of that 50 volt part. Seems to me it'll see nearly 100 volts V_RR ,

old jim

 

[yungman]

Actually I think I just use the name on the datasheet I downloaded. This is a universal footprint. The two I bought are either 400V or 1000V parts. I should have put DF10.

I want to show you the layout of the two rectifiers the smaller one within the bigger DF10 footprint so I can solder either one. Notice the GREEN outline indicates that the part is on the bottom side of the pcb.

This one show I have the optional J27 connector inside the footprint of the transformer so if I don't use the transformer, I can connect the AC of the main transformer to power the circuit.

 

[yungman]

If the transformer secondaries are independent, can you build a pair of identical floating "positive" DC supplies then connect them so the box presents (+) (0) (–) at its output?

Sorry I did not see your post until now. This is the transformer I use https://www.mouser.com/ProductDetail/530-LP-56-100

I drew it out using one of the secondary winding:

Will this work? Seems like so. Ha ha, it would be too easy. I would save a lot of diodes.

Thanks

 

[yungman]

Here is the updated schematic.

 

[yungman]

I have one more question. I need to keep Vopsa and Vipsa to within 6V( -Vopas and -Vipsa also). I use a pair of darlingtons Q1 and Q2 to do that as shown. If you have any better idea, let me know. The idea is when power off, Vopsa is going to collapse a lot faster, when Vopsa drop to 4.7V below Vipas, the 4.7V zener will turn on and start developing voltage across R3. when the voltage reach about 1.3V, Q1 will turn on and clamp the voltage to within 6V. D2 serves as the other way and to protect Q1.

Here are the info for the two darlingtons
https://www.digikey.com/product-detail/en/on-semiconductor/BDV65BG/BDV65BGOS-ND/1476194

https://www.digikey.com/product-detail/en/on-semiconductor/BDV64BG/BDV64BGOS-ND/1476192

There is not resistor from the driver transistor to the emitter of the main transistor in the darlington of these two transistor, this is perfect.

 

[eq1]

I haven't had time to grok the full schematic, but I can give a quick critique of the circuits in post #15. This is without knowing what the load on J1 & J6 is, but I'll assume they can be some low ohmic load since you needed a Darlington to hold it up and the current that leaves J1 returns into J6.

• I suspect you're trying to regulate it to 6V away and not clamp it (I read the wording in post #15 as a clamping problem). If it's a clamp, can you short them? 0V is within 6V. :)
• I think the D12, Q1 feedback circuit is a simple and elegant way to regulate the inputs to 6V away. One question though, can you drop 12V on the Q1 circuit and delete the Q2 circuit? Maybe you split them for thermal reasons? You could put a zener in series with the collector on Q1 to split the power on another component if that was the case. I know the board is made but with one circuit you don't have to worry about what happens when the two are not symmetrical, for example, what if they have different bandwidths?
  
• The tolerance on CMDZ5230 is pretty high especially as there is already ~15mA going through it due to R3. Vz varies over ~500mV which is going to put just the DC regulation accuracy at over 10%. Maybe that's within spec though.
• You should consider a small emitter resistor on the Darlingtons. (Perhaps the load itself is the emitter resistance?) A resistor will limit the turn-on pulse (presently it shoots straight through D12 and Q1), reduce the risk of thermal runaway, and with C48 give more precise control over the circuit's bandwidth (by limiting it).

Just $0.02. I put a lot of comments but most likely the circuit is just fine unmodified since J1/J6 probably power something basic like a tube heater or something (I'm guessing based off your previous posts).

 

[yungman]

The schematic is very confusing, this is the simplified version. The power amp has 4 separated power supplies, 2 for each channel. I have a =/-35V supplies to power up the frontend(+/-Vipsa) of the power amp, +/-33V to power just the output section(+/-Vopsa). If you see the drawing, I have a lot of filter caps. 50,000uF for each rail of the output section, 15,000uF for each rail of the frontend section. To protect the circuit, the voltages of each the frontend and the output section has to track to within say 6V. I drew diodes and zener diodes between the two power supplies so when one collapses, it will drag the other down to keep the differential within the safe limit.

I want the front end to have higher voltage to lower distortion at high signal level, I don't want both sides to have the same voltages.

The transistor with zener diode in the circuit is just substitution of that 6V zener of this drawing. I cannot use zener, I tried as explain before.