A very nice power supply with an smps pre regulator and a linear regulator output

[William Gaatjes]

The DC2132A is a very nice powersupply with specs 0-24V and 0 to 3A. And the nice part is that it does not require any heatsinks. It has a step down smps that pre-regulates the voltage down to Voutput + 1,7V. And behind the smps is a linear regulated output stage (comprised of a dual LT3081 setup) to remove any ripple to get a clean output voltage. Added features are thermal monitoring and current monitoring, courtesy of the LT3081.
The LT3081 is used in a dual setup to allow a maximum current of 3A.

Oh well, how the circuit works, a LTspice model is available as is a demoboard, which i am so lucky to posses. Was sort of needed for work shortly and after it was no longer needed, i could have it. ^_^.

http://www.linear.com/solutions/5086
LT journal with explanation.
http://www.linear.com/docs/45093

As i played with it at home, i found out a worrying fact, when the current control is turned to zero, the output would give of a negative voltage. -0.76V to be exact.
Parasitic diodes.
The problem is that most integrated circuits do not handle negative voltages on their powersupply pins. This is because how ic are build up, by silicon and P and N layers. All ic's have some form of parasitic diodes and parasitic thyristors because of the way the silicon is doped . Usually, we never notice because we apply the correct voltages. and in all datasheets the absolute ratings of supply voltages are given. For example -0.3V to 6V.
https://en.wikipedia.org/wiki/Parasitic_structure

Solution :
So, i had to find a way to solve this. The whole point is that the LT3081 needs a load current with a maximum of 5mA to function. But how are you going to let this current flow if the output is 0V with respect to ground level. Well, a negative powersupply of -5V is used. This -5V is created with another smps. Since it is a regulated voltage, all that is needed is a resistor R2 and a transistor Q1. It might just be that this is the case with my pcb or that it is systematic, anyway this way the bench powersupply is useless. And the problem is because the R2 and Q1 are a static circuit. If there is any tolerance, this circuit will not work. At least, not with my pcb.

My first attempt was to replace R2 with a 470Ohm resistor and a 500 Ohm potentiometer. But when i calibrated the output to 0V, i noticed something problematic. When i quickly turned the voltage potentiometer up and down again, the output voltage would start to float at around 9V. Even while the set input of the LT3081 is at 0V. Only when i turned the voltage potentio meter slowly, would the LT3081 output follow the voltage at the set input.
This was useless and unacceptable and had to think of something else.
Since i am going to use a 10 turn potentiometer for voltage and current control anyway, this might not be an issue but i want to be able to trust my bench power supply.

So, i removed R2 and Q1 and made a circuit with an LM358 (temporary solution, i am going to use another) opamp as an integrator, a transistor with emitter resistor to sink the 2x 5mA(maximum for 2x LT3081) current dynamically instead of statically. I will make a schematic later and post it. But anyway, the inverting input i connected to ground through a 10K resistor and the non inverting input through a 10k resistor to the output. And added a diode at the non inverting terminal of the LM358, so that the maximum common mode voltage of the opamp is never exceeded(output can be 24V). The non inverting input will never come higher then 0.6V with respect to ground this way.

The opamp will try to keep the output at exactly 0V. But it can only sink about 10mA. That is no match for the LT3081 combination that sources a maximum of 3A. But when the output voltage is reduced to zero and the current control is reduced to zero, the opamp circuit i made comes in effect, keeping the load current through the LT3081 combination flowing at a value to make sure that the output voltage is then 0V. And that the load current is high enough that the two LT3081 acutally see a load current since they need this current to function properly.
In the LT journal it is all explained how the circuit works. I will post a schematic later on about the opamp circuit i added.

Why i do this ?
The plan is to add a PIC16LF1509 to read out the voltages that represent the output current and the output voltage. also, because the current is available as a voltage and i always like to have a current trip feature next to a current source mode feature. When i set the current potentiometer to maximum and i set the pic in current trip mode with an extra potentiometer, i can set the trip current where the bench supply will be turned off by the pic.
A fuse mode. To protect expensive ic's in expensive proto circuits.
The pic will be connected to a 2x16 character LCD and a 2 buttons will be added. One for menu, and one to turn the powersupply output on and off.
And the pic will do thermal monitoring. When the temperature outputs of the LT3081(can be put in parallel since they are high impedance current sources) give a high enough voltage over an external resistor, for example the LT3081 has reached 100C, the pic will display "temperature over limit" and turn of the bench supply to protect the LT3081 from overheating.
I will then have a very luxurious bench power supply in a small case.

Schematic :

I do find it sad, that the evaluationboard of Linear Technology is a bit expensive. But the design files and the schematics are available to build one yourself.

I will post a schematic tomorrow about the current sink i created to get the output to a proper 0V.

LT3081 datasheet :
http://www.linear.com/docs/43256

 

[john5677]

The DC2132A is a very nice powersupply with specs 0-24V and 0 to 3A. And the nice part is that it does not require any heatsinks. It has a step down smps that pre-regulates the voltage down to Voutput + 1,7V. And behind the smps is a linear regulated output stage (comprised of a dual LT3081 setup) to remove any ripple to get a clean output voltage. Added features are thermal monitoring and current monitoring, courtesy of the LT3081.
The LT3081 is used in a dual setup to allow a maximum current of 3A.

Oh well, how the circuit works, a LTspice model is available as is a demoboard, which i am so lucky to posses. Was sort of needed for work shortly and after it was no longer needed, i could have it. ^_^.

http://www.linear.com/solutions/5086
LT journal with explanation.
http://www.linear.com/docs/45093

As i played with it at home, i found out a worrying fact, when the current control is turned to zero, the output would give of a negative voltage. -0.76V to be exact.
Parasitic diodes.
The problem is that most integrated circuits do not handle negative voltages on their powersupply pins. This is because how ic are build up, by silicon and P and N layers. All ic's have some form of parasitic diodes and parasitic thyristors because of the way the silicon is doped . Usually, we never notice because we apply the correct voltages. and in all datasheets the absolute ratings of supply voltages are given. For example -0.3V to 6V.
https://en.wikipedia.org/wiki/Parasitic_structure

Solution :
So, i had to find a way to solve this. The whole point is that the LT3081 needs a load current with a maximum of 5mA to function. But how are you going to let this current flow if the output is 0V with respect to ground level. Well, a negative powersupply of -5V is used. This -5V is created with another smps. Since it is a regulated voltage, all that is needed is a resistor R2 and a transistor Q1. It might just be that this is the case with my pcb or that it is systematic, anyway this way the bench powersupply is useless. And the problem is because the R2 and Q1 are a static circuit. If there is any tolerance, this circuit will not work. At least, not with my pcb.

My first attempt was to replace R2 with a 470Ohm resistor and a 500 Ohm potentiometer. But when i calibrated the output to 0V, i noticed something problematic. When i quickly turned the voltage potentiometer up and down again, the output voltage would start to float at around 9V. Even while the set input of the LT3081 is at 0V. Only when i turned the voltage potentio meter slowly, would the LT3081 output follow the voltage at the set input.
This was useless and unacceptable and had to think of something else.
Since i am going to use a 10 turn potentiometer for voltage and current control anyway, this might not be an issue but i want to be able to trust my bench power supply.

So, i made a circuit with an LM358 (temporary solution, i am going to use another) opamp as an integrator, a transistor with emitter resistor to sink the 2x 5mA(maximum for 2x LT3081) current dynamically instead of statically. I will make a schematic later and post it. But anyway, the inverting input i connected to ground through a 10K resistor and the non inverting input through a 10k resistor to the output. And added a diode at the non inverting terminal of the LM358, so that the maximum common mode voltage of the opamp is never exceeded(output can be 24V). The non inverting input will never come higher then 0.6V with respect to ground this way.

The opamp will try to keep the output at exactly 0V. But it can only sink about 10mA. That is no match for the LT3081 combination that sources a maximum of 3A. But when the output voltage is reduced to zero and the current control is reduced to zero, the opamp circuit i made comes in effect, keeping the load current through the LT3081 combination flowing at a value to make sure that the output voltage is then 0V. And that the load current is high enough that the two LT3081 acutally see a load current since they need this current to function properly.
In the LT journal it is all explained how the circuit works. I will post a schematic later on about the opamp circuit i added.

Why i do this ?
The plan is to add a PIC16LF1509 to read out the voltages that represent the output current and the output voltage. also, because the current is available as a voltage and i always like to have a current trip feature next to a current source mode feature. When i set the current potentiometer to maximum and i set the pic in current trip mode with an extra potentiometer, i can set the trip current where the bench supply will be turned off by the pic.
A fuse mode. To protect expensive ic's in expensive proto circuits.
The pic will be connected to a 2x16 character LCD and a 2 buttons will be added. One for menu, and one to turn the powersupply output on and off.
And the pic will do thermal monitoring. When the temperature outputs of the LT3081(can be put in parallel since they are high impedance current sources) give a high enough voltage over an external resistor, for example the LT3081 has reached 100C, the pic will display "temperature over limit" and turn of the bench supply to protect the LT3081 from overheating.
I will then have a very luxurious bench power supply in a small case.

Schematic :

I do find it sad, that the evaluationboard of Linear Technology is a bit expensive. But the design files and the schematics are available to build one yourself.

I will post a schematic tomorrow about the current sink i created to get the output to a proper 0V.

LT3081 datasheet :
http://www.linear.com/docs/43256

Your observations contradict the datasheet for DC2132A. "DC2132A operates to 0V and 0A."

 

[William Gaatjes]

Your observations contradict the datasheet for DC2132A. "DC2132A operates to 0V and 0A."

I know. Either this is a one time problem, only with this board or it is not. I do not know. I only have one example. I also did not feel like changing all components. I think it is tolerance related. Anyway, it is a too great bench supply to skip.

So, i solved the issue.

Do note, datasheets do not always tell every situation. At work we had an ic once where there was supposed to be no offset in the current monitoring section of the ic. But the ic has two current monitoring opamps. One for low side current measurement and one for high side current measurement. Now the low side opamp has ridiculously low offset, almost a perfect opamp. The high side opamp had offset unfortunately and it was not constant either. It was different with every ic we encountered during production of the circuit where the ic was in designed. We had to include offset calibration in the software to calibrate the offset to a minimum. And the datasheet gave the idea that there was only very low offset (in the uV).

 

[William Gaatjes]

Here is the schematic :

I should note that the transistor i use is a TO-92 casing.
It should be able to withstand the dissipation at Vout = 24V, although i still have to confirm that.

Some pictures of the prototyping stage on a breadboard.
I have been going to the gym again, taking boxing lessons and fitness workouts, so my hands are a bit shaky. Which is obvious to see from the pictures. I hope this side effect will go away because it is annoying.
Shaky hands is not productive for hobby or work at all. When i have to build up a prototype pcb (and i have to do now with 0402 components) i have difficulty keeping my hands steady while placing the components with a pincet.

 

[William Gaatjes]

I should note that i am going to remove the green led from the INTVCC supply of the LT8612. The current not flowing through the led, i will then have available for the sink current circuit. Although the circuit will consume very little current from the INTVCC 3,3V supply. The hfe(beta) of the transistor is at least 100x ,so the base current is 100uA max with 10mA load current, add the quiesient current for the planned to use opamp (LT1490A) and it would still be very well within the current limits of the INTVCC pin. The circuit draws less current than the green led.

 

[SOFTengCOMPelec]

That's an interesting, fun and very useful bench power supply project.

I'm confused by the schematic, you have supplied, with your? op-amp, output voltage NOT reaching 0 volts, fix.

Because (from the applications note schematic) transistor Q5, via a 549 Ohm resistor, connected to the -5V at 8 mA (via another, small SMPS). Is suppose to provide the necessary load current, to make it back to 0 Volts output voltage (minimum).

I.e. You should NOT need it, as Q5 + 549 Ohm resistor, should fix it, already. Maybe these were optional components, and have not been fitted, or something ?

Without any correction circuits at all. The datasheet seems to indicate approximately 700 milli volts of output voltage (but you seem to be talking about NEGATIVE voltages, see graphs at bottom of data sheet), when set voltage = 0 Volts and load = 0 Amps.

 

[William Gaatjes]

That's an interesting, fun and very useful bench power supply project.

I'm confused by the schematic, you have supplied, with your? op-amp, output voltage NOT reaching 0 volts, fix.

Because (from the applications note schematic) transistor Q5, via a 549 Ohm resistor, connected to the -5V at 8 mA (via another, small SMPS). Is suppose to provide the necessary load current, to make it back to 0 Volts output voltage (minimum).

I.e. You should NOT need it, as Q5 + 549 Ohm resistor, should fix it, already. Maybe these were optional components, and have not been fitted, or something ?

Without any correction circuits at all. The datasheet seems to indicate approximately 700 milli volts of output voltage (but you seem to be talking about NEGATIVE voltages, see graphs at bottom of data sheet), when set voltage = 0 Volts and load = 0 Amps.

The whole point is that the transistor Q1 and resistor R2 are providing to much current at the wrong time, causing the output to not be 0V at 0V setting but to be -0.76V. And that i do not want. It is a very critical part of the DC2132A circuit.
Hence the circuit i made that actively tries to compensate for 0V at 0V setting while at the same time letting enough load current flow that the LT3081 combo actually can function. As long as there is an external load and the output is above 0V, there is no problem because the external load will sink the current. Of course, i have removed R2 and Q5. Since there is no use for it anymore. I replaced with an active current sink circuit.

 

[SOFTengCOMPelec]

Redacted.

(It's problematic, to try to help people with potentially complex stuff, in forums. Sorry.)

 

[William Gaatjes]

Redacted.

(It's problematic, to try to help people with potentially complex stuff, in forums. Sorry.)

Redacted ? Buddy, what do you mean ?
We are electronic enthusiasts, people like us always argue, that is how we learn. habit of the interest/hobby. I think you are great. You are my pal !

 

[SOFTengCOMPelec]

Redacted ? Buddy, what do you mean ?
We are electronic enthusiasts, people like us always argue, that is how we learn. habit of the interest/hobby. I think you are great. You are my pal !

Un-Redacted, and replaced, here:

The whole point is that the transistor Q1 and resistor R2 are providing to much current at the wrong time, causing the output to not be 0V at 0V setting but to be -0.76V. And that i do not want. It is a very critical part of the DC2132A circuit.
Hence the circuit i made that actively tries to compensate for 0V at 0V setting while at the same time letting enough load current flow that the LT3081 combo actually can function. As long as there is an external load and the output is above 0V, there is no problem because the external load will sink the current. Of course, i have removed R2 and Q5. Since there is no use for it anymore. I replaced with an active current sink circuit.

It seems very odd, that you are needing to design and build a correction circuit. For something which should already be designed/tested and correctly working, assuming functional parts and correct assembly.

EDIT: When I look at their circuit, it seems to show that the minimum current setting possible (50 .. 100 mA, very approximately), would easily exceed the 8 mA from the bits you removed.

Assuming you stuck to their plans, when you wired in the current controls.

EDIT2:
I've seen small negative voltages, similar to what you describe when you turn the current knob, fully off, on commercial power supplies. If I remember correctly.
It annoys me. E.g. Germanium transistors, with lower voltages, could be damaged, maybe.
Also I hate giving polarized capacitors, voltages in the WRONG direction.

 

[SOFTengCOMPelec]

On further thought. Maybe you are getting confused, because the (usual) Volts and Amps, displays/meters are currently MISSING.

On some power supplies (if I remember correctly). If you set them to show, say 12.000 Voilts. Then turn the current limit TOO far to the left (on some ONLY, depending on how it is set up). The voltage (12.000V) is seen to collapse.

I.e. You would tend to remember, to leave a bit of play, on the current limit knob. Hence 0.000V will be 0 Volts.

The above is just guess work, and could easily be wrong.

 

[William Gaatjes]

Un-Redacted, and replaced, here:



It seems very odd, that you are needing to design and build a correction circuit. For something which should already be designed/tested and correctly working, assuming functional parts and correct assembly.

EDIT: When I look at their circuit, it seems to show that the minimum current setting possible (50 .. 100 mA, very approximately), would easily exceed the 8 mA from the bits you removed.

Assuming you stuck to their plans, when you wired in the current controls.

EDIT2:
I've seen small negative voltages, similar to what you describe when you turn the current knob, fully off, on commercial power supplies. If I remember correctly.
It annoys me. E.g. Germanium transistors, with lower voltages, could be damaged, maybe.
Also I hate giving polarized capacitors, voltages in the WRONG direction.

Ah, there is perhaps the crux. I do really want to be able to have a minimum current setting of about 10mA. When i have 10mA , i am happy because i can test a led with it. But i really would like to have a bench supply where i could turn the current knob all the way to zero without side effects.

 

[SOFTengCOMPelec]

If their original circuit, can/does indeed, produce the -0.76 volts. I completely agree with you NOT liking that (I dislike it myself), and fixing it. E.g. With the op-amp.

But be careful, that the circuit does NOT misbehave. I've had problems with some power supplies, because the back-emf and/or interference from motors/relays etc, has upset the power supply.

Good quality ones, are usually unaffected by such stuff.

 

[William Gaatjes]

Well, i do have to say that i am going to add a schottky diode in reverse to the terminals to protect the bench power supply from inductive loads generating negative spikes on the VOUT terminal with respect to ground. The negative voltage should never be any lower than -0.3V. Schottky diodes have a lower forward voltage depending on the forward current but can be easily near 1V depending on the (very high) forward current. But i really would like to have 0V.

 

[William Gaatjes]

That is fun, had the idea to post about a reverse diode without reading your post (#13) yet. Great minds think alike .

 

[SOFTengCOMPelec]

Ah, there is perhaps the crux. I do really want to be able to have a minimum current setting of about 10mA. When i have 10mA , i am happy because i can test a led with it. But i really would like to have a bench supply where i could turn the current knob all the way to zero without side effects.

I know exactly what you mean.

I somewhat HATE power supplies, where the voltage meter reading collapses (when the current limit is set to zero and the output isolator switch is open/off). As it makes setting the voltage a pain in the neck.

I like to test new/fresh circuits, with very low initial currents. To minimize/eliminate damage, if I have mis-connected some stuff.

I guess power supplies are a bunch of compromises. However it is set up, there will always be times, when that way is problematic.

 

[William Gaatjes]

I do have to write, that remembering various power supplies, that turning the current knob to 0A, always lowers the voltage to 0V. And that is oke with me. But i do not want the output to be negative because then parasitic diodes and thyristors in ic's and circuits could become active.

 

[john5677]

Where did you obtain the evaluation board? Did someone use it before you? I have a good idea where the problem lies on the power supply.

 

[William Gaatjes]

Where did you obtain the evaluation board? Did someone use it before you? I have a good idea where the problem lies on the power supply.

It was brand new from LT itself for as far as i know. But i got it through a distributor. So who knows. I am not going to blame anyone. My main objective with this thread was to let everybody know about a nice bench power supply with a lot of neat tricks to keep dissipation at bay while still being a very good supply with good transient response with hardly any capacitance.

Anyway it works great now. All i have to do now is to find a small form factor 230V to 24V 3A SMPS which i can adjust up to about 29V(I hope that is enough). And i have to make a small prototype pcb with a switching regulator(NCP3063, DIL8 case, easy soldering) ~24V to 4V to power a linear 3V3 regulator(MCP1700-3300) to power a PIC. It will be great.

http://www.onsemi.com/pub/Collateral/NCP3063-D.PDF
http://www.microchip.com/wwwproducts/Devices.aspx?product=MCP1700

A side question, what is your take on it ?

 

[john5677]

It was brand new from LT itself for as far as i know. But i got it through a distributor. So who knows. I am not going to blame anyone. My main objective with this thread was to let everybody know about a nice bench power supply with a lot of neat tricks to keep dissipation at bay while still being a very good supply with good transient response with hardly any capacitance.

Anyway it works great now. All i have to do now is to find a small form factor 230V to 24V 3A SMPS which i can adjust up to about 29V(I hope that is enough). And i have to make a small prototype pcb with a switching regulator(NCP3063) ~24V to 4V to power a linear 3V3 regulator to power a PIC. It will be great.

A side question, what is your take on it ?

Did someone use the demo board before you?

 

[William Gaatjes]

Did someone use the demo board before you?

For as far as i can tell, it was brand new. And even then, i could just have a bad version, it rarely happens but it does happen. But i fixed it and it works. I am happy with it. Monday i will be desoldering the potentiometers and replacing them with 10 turn potentiometers.
Even in the almost impossible case, it is a design flaw, i do not mind. It works now.
But i do find it strange, there is no reverse diode on the output.

 

[SOFTengCOMPelec]

For as far as i can tell, it was brand new. And even then, i could just have a bad version, it rarely happens but it does happen. But i fixed it and it works. I am happy with it. Monday i will be desoldering the potentiometers and replacing them with 10 turn potentiometers.
Even in the almost impossible case, it is a design flaw, i do not mind. It works now.
But i do find it strange, there is no reverse diode on the output.

Apparently it HAS already got reverse output voltage protection. Page 21 of the datasheet mentions it. Along with many other built in protection devices.

 

[William Gaatjes]

Did someone use the demo board before you?

Not at my work, and i did not damage it either. For as far as i know it is brand new.

 

[William Gaatjes]

Apparently it HAS already got reverse output voltage protection. Page 21 of the datasheet mentions it. Along with many other built in protection devices.

Yeah, i did read about that in the LT3081 datasheet. But i just want it to be as robust as possible. It would be sad if the emf voltage spike of a coil or a motor (inductive load) would damage the psu. An extra schottky diode is a good idea IMHO.

 

[SOFTengCOMPelec]

Yeah, i did read about that in the LT3081 datasheet. But i just want it to be as robust as possible. It would be sad if the emf voltage spike of a coil or a motor (inductive load) would damage the psu. An extra schottky diode is a good idea IMHO.

I agree, that makes sense.

Similar example:
Integrated circuits, often/usually claim they have built in, anti-static damage, protection diodes.

But good circuit designers, often/sometimes add their own protection, diodes and resistors etc. Between the chip(s) and the outside world.

It is fairly well known, that the in-built protection diodes, can only take a limited amount of abuse (energy). Then potentially the device can be damaged. Properly designed (and protected), input(s), is often the best way to go.

But NOT everyone, would agree with this.