0-400V digital PSU design

[lordsandman]

Hi,
i want to design digitally controlled 0-400V DC power supply with a floating output. max output current may 2mA or a bit more. the negative output of the PSU will connect to another equivalent PSU output but electrically isolated.
when i searched about this concept a few ideas occurred. first; it might be a linear voltage regulator which have a reference voltage is derived from a DAC. Second; a DAC and opamp(for virtual ground) based voltage divider can divide a regulated PS output. so its an improved digital potentiometer practically.
i cannot decide which of these design works best. target output performance like this: <0.4mV rms ripple, <50ppm temperature and <100ppm voltage drift.

i need help and open to other suggestions

 

[ronv]

I think you will have trouble with the 50 ppm temperature and 100 ppm drift.

What will you be using as the "bulk supply" just a transformer?

Here is a linear that would work, but as you can see in the data sheet it is 1% over its temperature range.

https://www.electro-tech-online.com/custompdfs/2012/03/LR8.pdf

 

[alec_t]

I second Ronv's doubts on drift. What are you doing that requires that accuracy?

 

[lordsandman]

hello
i heard about LR8 but i suspect about such a tiny device works stable as i desire. i'll use this power supply to create electrostatic field between metal plates. i cited the drift values from a similar project.
here is the scheme and output performance table:

View attachment 62369

View attachment 62370

i dont use the same circuit because it needs seperate dc supply for every controlled output.

 

[Nigel Goodwin]

The diagram doesn't seem to make such sense?, and the opamp will disappear in a big puff of smoke.

 

[lordsandman]

ok. how can i achive this? what kind of circuit i must use?
i have to work with a minimum ripple, say 10mV maximum. could have done without excellent temperature drifts. there are commercial HV opamps like Apex PA85 and modules like Emco CA series but unit price starts from 200$. i'm going to build 20 independent identical modules. above technique is a plan for now and i see that its useless. i would appreciate if you help about designing

how about this??

View attachment 62378

 

[lordsandman]

The diagram doesn't seem to make such sense?, and the opamp will disappear in a big puff of smoke.

in the diagram opamp's bipolar supply grounded at output voltage. so, opamp input between 0-10V

 

[alec_t]

The post #6 circuit looks more promising. I guess the devil will be in the detail if you want high accuracy and low drift.

 

[Nigel Goodwin]

in the diagram opamp's bipolar supply grounded at output voltage. so, opamp input between 0-10V

Your diagram didn't show that!.

 

[ronv]

Way back you say floating. Talk some more about that.

 

[lordsandman]

what kind of devilness should i commit?
at first sight (post #6 circuit) i can see; the opamp must have low temperature drift, offset voltage and input bias current. a series output resistor and protection diodes may be necessary, together with capacitors to reduce noise and ripple. electrolytic capacitors must be avoided i guess, as they have leakage currents that are often significant when compared with FET source current. also, voltage reference has a critical effect.

an output circuit like this can work properly?
View attachment 62429

Way back you say floating. Talk some more about that.

i'll connect one PSU outputs (+ or -) to any part of another circuit or other PSU and not cause a short. i draw a scheme, PSU1 positive output connected to earth. PSU2 and 3 outputs referred to PSU1 and PSU4 referred to PSU3, and so on.

View attachment 62430

this is what i mean from floating output

 

[alec_t]

an output circuit like this can work properly?

If the output voltage is 400V the 10k resistor (R11) will have to dissipate 16W. Most unlikely to get a stable output then. Also, as the circuit in post #6 is essentially current-controlled Vs (400V) will have to be very stable. Not clear how you plan to achieve that.

 

[ronv]

High voltage supply

The good news is that you don't need big filters because the output current is so low (2 ma)

Current limit is about 4 ma.

Take a look at this circuit. The transistors are only good for 450 volts so you need to watch your input supply. I just used a bridge so I could look at ripple. (3 mv)

I think the only thing you need to worry about is your DAC reference as far as drift goes.

Might want a small heatsink as the power at low voltage is about a watt.

 

[lordsandman]

alec_t and ronv, thank you so much for your attention. and sorry for my bad english by the way. i'm not well experienced in analog electronics so i dobt about the design and indecisive about this. because i'm a digital guy and working progress in analog. i'm learning a lot in this project. the first thing i've learned is isis a very bad simulator in analog electronics. ronv; very impressed in spice simulations, once again. i guess its time to learn spice derivatives. i strived with multisim and tina whole day as a beginner step

and i can not decide to technique yet. i think i have 2 options: 1-current control with DAC 2-linear regulator based one
in your personal point of view which one is the best choice for me? or can you suggest another method?


EDIT:

If the output voltage is 400V the 10k resistor (R11) will have to dissipate 16W. Most unlikely to get a stable output then. Also, as the circuit in post #6 is essentially current-controlled Vs (400V) will have to be very stable. Not clear how you plan to achieve that.

10k resistor gettin there by mistake as default value of isis. i planned that a series of resistor instead of this

and ronv, can you send the spice file to me? so i can play with it

 

[ronv]

Attached is the file for the regulator. Just for fun I ran the temperature test. Not to bad I guess. I think most the temperature drift is due to R4 & R6 so it could be made better with 25ppm resistors.
I made a couple of changes. I'm still not real happy with the stability.
What are you making that you are looking for such tight tolerances?
Don't forget if you use a DAC like this design assumes each step will be about .3 volts.
The op amp and transistors are not in the standard spice library but for experiments you can replace them with any npn transistor and rail to rail op amps. (It will let you run things way out of spec.) If you are just getting started downloading and installing new models will make you unhappy with spice.

 

[lordsandman]

Attached is the file for the regulator. Just for fun I ran the temperature test. Not to bad I guess. I think most the temperature drift is due to R4 & R6 so it could be made better with 25ppm resistors.
I made a couple of changes. I'm still not real happy with the stability.
What are you making that you are looking for such tight tolerances?
Don't forget if you use a DAC like this design assumes each step will be about .3 volts.
The op amp and transistors are not in the standard spice library but for experiments you can replace them with any npn transistor and rail to rail op amps. (It will let you run things way out of spec.) If you are just getting started downloading and installing new models will make you unhappy with spice.

i'll use this power supply to create electrostatic field between metal plates. there are ions in the enviroment, so the only load current is the ions which hit the plates. PSU must be very stable and ripple free.
thanks for the spice schem, but when i simulate it output voltage decreases after 2 second and goes to zero after 20s. i changed opamp with universal opamp and transistor with 2n2222. why is this happening?

 

[ronv]

The voltage drops because I only put in 100 cycles for the ac power source on the far left hand side. You can right click the voltage symbol and change the number of cycles. If you change c2 to 1000pf and C1 to .1 ufd. the circuit will be more stable and have less ripple.

 

[lordsandman]

ohh! accept as my verdancy. but i have another problem. when i changed input AC voltage(V1) amplitude and frequency ripple changes dramatically. do transformer parameters cause this?

 

[ronv]

The transformer is just an ordanary power supply so the output of the transformer will follow the input. For example if the input voltage goes down 10% the output (Vin) will also go down. The ripple at the output of the bridge rectifier will go up if for example if you change from 60 Hz to 50 Hz but the output of the regulator will remain steady unless Vin gets to low or the frequency gets to low for the filter ( C3 ) to smooth it out. I didn't see any change in the output in either case.
What is your input to the regulator?

 

[lordsandman]

ronv, thanks for your kindly help. i found my mistake its working now when i was playing with analysis commands some parameters kept false. i like LTspice, i'm going to use this in analog simulation. it'll be excellent if it were not restricted in Linear devices, but its still wonderful for beginners like me.

is this circuit works for long times in real world? cause transistors makes me uneasy. i used 2N5550 npn transistors (Vceo=150V) in simulation. what am i suppose to use in real circuit and is it necessary to cool them? and i can't figured out the purpose of Q1.
how can i add a current limiting function to the design? connecting a current latch IC to a sense resistor which cuts off the Q3 is valid? it would be perfect because i can get an alarm if overcurrent exists.

finally, when i connect one PS's output to another's output its gone crazy. isn't it suppose to be float?