Is 550uF too big for a power supply that has CC limit?

[Doctorandus_P]

It's all but impossible to design a PSU with constant current and constant voltage and have good load regulation for both. A capacitor will improve the transient response of the CV mode, at the expense of CC. No capacitor an a large inductor will improve the CC mode, at the expense of CV mode.

Could it be a good idea to put a MOSfet in series with the ouput capacitor and disconnect it in CC mode?

Another point of attention are the sense lines. If these are connected to the outside world to compensate voltage drop over external wiring they can get attached to external voltages and this may lead to excessive current through the 1n4148 diodes.
If Vsense_Pos gets grounded somehow it looks like your power supply wil output the maximum voltage it's capable of. Owtch.

[bloguetronica]

I think that this other power supply, that has the corresponding schematic attached, might have stability issues too. I ran into oscillations caused by a  misplaced ground connection (www.eevblog.com/forum/projects/usb-controlled-precision-power-supply-(or-voltage-reference)/msg2007782/#msg2007782). Normally, this supply is stable, but the constant current load that that you see is being supplied via a SMPS and causes some oscillation, as long as it is not grounded to earth.

Maybe I am reading the schematic wrong but it looks to me like the remote sense to the voltage feedback loop is backwards, but it reminds me of a circuit I was studying a couple weeks ago from the LT1010 datasheet shown below.  The OPA703 is slow enough that it should be able to control the output through a TIP31 emitter follower without problems.

The first example shown below is a good example of what I was talking about.  D2 clamps the current control operational amplifier to prevent integrator windup.  C1 and to a lessor extend C2 maintain stability despite the current control loop being in series with the voltage control loop with two relatively fast operational amplifiers.  This design requires no output capacitance but has what is effectively a fast full class-ab output stage in the LT1010.  Note that the 2 ohm resistor in series with the output of the power stage aids stability when a capacitive load is present and this is an advantage of having the current shunt in series with the output.

The second example is from National Semiconductor is the low output capacitance design I was thinking of.  D2 takes advantage of the external compensation feature of the LM301A current control operational amplifier to clamp it preventing integrator windup preserving fast response.  Q2 makes the output stage more like class-ab with the ability to pull the output down.

The third example uses 100 microfarads of output capacitance for a 1/2 amp output current to control transient response because no clamping of the operational amplifiers is used.  In practice I think that capacitance could be 22 microfarads without problems but they wanted extra stability.

The remote sense is not backwards, that much I can guarantee. The resistor networks that you see are a means to provide internal lead compensationfor both the positive and ground leads. The schematics you've shown seem to be completely different applications.

...
Could it be a good idea to put a MOSfet in series with the ouput capacitor and disconnect it in CC mode?
...

No, that would actually cause instability. You don't want a capacitor connecting and disconnesting when the supply is constantly jumping between CV and CC modes.

...
Another point of attention are the sense lines. If these are connected to the outside world to compensate voltage drop over external wiring they can get attached to external voltages and this may lead to excessive current through the 1n4148 diodes.
If Vsense_Pos gets grounded somehow it looks like your power supply wil output the maximum voltage it's capable of. Owtch.

Any supply with remote sensing is capable to do that. Some even will output the maximum when the sense wires are not connected. This one will output 10% more of the intended voltage in this case. However, I can't guarantee cases of bad usage (the cases you mentioned). The diodes are there just for EDS protection.

Kind regards, Samuel Lourenço

[David Hess]

...
Another point of attention are the sense lines. If these are connected to the outside world to compensate voltage drop over external wiring they can get attached to external voltages and this may lead to excessive current through the 1n4148 diodes.
If Vsense_Pos gets grounded somehow it looks like your power supply wil output the maximum voltage it's capable of. Owtch.

Any supply with remote sensing is capable to do that. Some even will output the maximum when the sense wires are not connected. This one will output 10% more of the intended voltage in this case. However, I can't guarantee cases of bad usage (the cases you mentioned). The diodes are there just for EDS protection.

If the sense inputs are relatively high impedance, then medium value resistors can be used between force and sense at the power supply to prevent failure if a sense line becomes disconnected.

[bloguetronica]

...
Another point of attention are the sense lines. If these are connected to the outside world to compensate voltage drop over external wiring they can get attached to external voltages and this may lead to excessive current through the 1n4148 diodes.
If Vsense_Pos gets grounded somehow it looks like your power supply wil output the maximum voltage it's capable of. Owtch.

Any supply with remote sensing is capable to do that. Some even will output the maximum when the sense wires are not connected. This one will output 10% more of the intended voltage in this case. However, I can't guarantee cases of bad usage (the cases you mentioned). The diodes are there just for EDS protection.

If the sense inputs are relatively high impedance, then medium value resistors can be used between force and sense at the power supply to prevent failure if a sense line becomes disconnected.

Indeed! And that is the purpose of R18 and R19. They were chosen as a sort of compromise. The output voltage will not get higher than 10% when the sense is disconnected on both sides. However, they don't have such a low value that may affect precision due to lead resistance (think of a shorted R18 and R19 that would defeat the purpose). They also wont dissipate too much current if the sense leads are switched.

However, if the sense leads are switched, or if one sense wire gets shorted to the opposite polarity, the load will probably be damaged despite the supply being not.

Kind regards, Samuel Lourenço

[bloguetronica]

I think that R5 might be causing some delay and, therefore, instability. I wonder if adding a capacitor in parallel will improve things. Notice that I can't simply short out R5 because I don't want to short the IC4B amplifier's output.

I employed a similar solution today to one of my supplies. See this post:
https://www.eevblog.com/forum/projects/improving-the-fau200-power-supply/

Of course, I'll probably have to add some capacitors elsewhere in the loop (near IC4B), as described in previous posts.

Kind regards, Samuel Lourenço

[bloguetronica]

So, I've been doing some mods to the FAU201 power supply too, which is somewhat similar to the one in the OP. Definitely, I'll have to put a capacitor between the output and the inverting input of the op-amp, if I want to use a capacitor in parallel with R5. The last capacitor (in parallel with R5) will increase the high frequency response, so the former is required in order to decrease the high frequency noise.

In the case of FAU201, a 10nF from the op-amp output to the inverting input would greatly reduce the HF noise, although it was only a slight improvement comparing to the effect of a 3.3nF cap in the same place.

Edit: Added some measurements before and after the mods. They are full of noise because of the cable that I've had to use. The scope probe doesn't fit well between the banana plugs. Even though, you can notice the improvement.

P.S.: Unshielded cables make good antennas

Kind regards, Samuel Lourenço

[xavier60]

I have learned much from experimenting with this circuit, https://dangerfromdeer.com/2016/04/06/bench-power-supply-build-part-ii/#jp-carousel-898
Notice that the author gets it totally wrong with the compensation.
After adding proper compensation and other changes, it is now very stable. I did some experimenting with the output capacitor last night.
CV mode is stable with a 1uF MLCC in series with 0.5Ω. CC mode is stable with no output capacitor because the output stage itself is a current source, controlled by the error amplifiers.
I have mostly solved the CC op-amp's windup problem. The CC op-amp is allowed to slew at its full speed  until it has taken control of the output stage.
The schematic is on this page, https://www.eevblog.com/forum/projects/linear-lab-power-supply/300/
I don't believe that it has turned into an over complicated monster yet.   
I recently changed the sharing resistors to 50mΩ, reducing dropout to 1.65V @ 4A.

BTW: the Darlington at the output is not an Emitter Follower. It is driven by the voltage drop across the 2.2K caused by the current sourced by the PNP transistor, Q1.

[bloguetronica]

I have learned much from experimenting with this circuit, https://dangerfromdeer.com/2016/04/06/bench-power-supply-build-part-ii/#jp-carousel-898
Notice that the author gets it totally wrong with the compensation.
After adding proper compensation and other changes, it is now very stable. I did some experimenting with the output capacitor last night.
CV mode is stable with a 1uF MLCC in series with 0.5Ω. CC mode is stable with no output capacitor because the output stage itself is a current source, controlled by the error amplifiers.
I have mostly solved the CC op-amp's windup problem. The CC op-amp is allowed to slew at its full speed  until it has taken control of the output stage.
The schematic is on this page, https://www.eevblog.com/forum/projects/linear-lab-power-supply/300/
I don't believe that it has turned into an over complicated monster yet.   
I recently changed the sharing resistors to 50mΩ, reducing dropout to 1.65V @ 4A.

BTW: the Darlington at the output is not an Emitter Follower. It is driven by the voltage drop across the 2.2K caused by the current sourced by the PNP transistor, Q1.

I must say that I've learned plenty in the last few days, while reading the comments, seeing some videos and making the mods. Definitely, increasing the speed and compensating (increasing phase margin) is the way to go.

I've took new measurements regarding the AC noise at the output of the FAU201 power supply when loaded in the same conditions. This time, I've used the scope probe and a short piece of wire. One of the measurements was done after the mod, the other before. The noise reduction is incredible. Both capacitors are instrumental to increase phase margin.

I'll apply the same principle to the project in the OP, in an effort to reduce the output capacitance and take out or reduce some capacitors that are masking the issue. Type I compensators (TI terminology) near IC4B and IC7 are definitely needed. Perhaps, I should use a type II compensator near the latter op-amp.

Thanks for all the help so far!

Kind regards, Samuel Lourenço

[xavier60]

My main point is, it is difficult to make Current Control work properly with a Voltage Follower output stage.

[chickenHeadKnob]

It's all but impossible to design a PSU with constant current and constant voltage and have good load regulation for both. A capacitor will improve the transient response of the CV mode, at the expense of CC. No capacitor an a large inductor will improve the CC mode, at the expense of CV mode.

Normally we optimise for CV mode and accept that current surges beyond the current limit can occur. I've toyed with the idea of building a CV/CC power supply with a three position switch which adjusts the filtering to be optimum for CC, CV or the best of both, because no PSU I've seen offers this, although I've never got round to it.

Well the HP 6632B and related family members have a fast/normal mode switch , located inconveniently on the back.  Quoted from the user manual:

Fast/Normal Operation

A switch on the rear of the unit lets you switch between operating in either Fast mode or Normal mode.
When set to Fast mode, this switch disconnects the output capacitor that is located inside the unit. Fast
mode lets you improve or enhance certain operating characteristics; while at the same time degrading
other operating characteristics.
1. In Fast mode, the programming time for voltage programming is faster than for normal operation,
however, output noise is greater.
2. In Fast mode, the absence of the internal output capacitor results in increased output impedance and
therefore, greater stability when driving inductive loads. Conversely, the addition of external
capacitive loads in Fast mode will reduce the stability of the unit during constant voltage operation.
3. In Normal mode, the internal output capacitor helps control peak voltage excursions away from the
the nominal value for sudden changes in load current. In Fast mode, larger peak voltage excursions
will show up at the output of the unit during sudden load current changes.

Capacitive Loading
In Normal mode, the dc source will be stable for many load capacitances, however, large load
capacitances may cause ringing in the dc source’s transient response. If this occurs, the problem may be
solved by increasing or decreasing the total load capacitance.

In Fast mode, the dc source can maintain stability only for small capacitive loads. These limits are:
Agilent 6631B  2.2 μF.
Agilent 6632B and 66322A 1.0 μF.
Agilent 6633B                     .22 μF.
Agilent 6634B                     0.10 μF.


Looking at the schematic I am lost but it appears the capacitance that is switched is 100 mu Farad for the 6632b, 50 or 22 for the other models
I think having such a feature enabled by a front panel switch on a hobby purpose built supply is not a bad idea, however it doesn't eliminate the need to design for minimum output capacitance though.

[bloguetronica]

My main point is, it is difficult to make Current Control work properly with a Voltage Follower output stage.

What do you mean? I think there is some confusion. IC7, the op-amp responsible for the CC regulation, is not a voltage follower. The circuit in the OP works, but is not very well tuned.

There are two DACs: IC3 sets the voltage and IC5 sets the current limit. The only voltage follower here is IC4A, and it is needed as the IC3 DAC buffer. IC4B is the error amplifier, IC6 amplifies the sensed voltage in order to "read" the current and IC7 acts as a comparator of sorts that sets the CC mode when the current is exceeded.

Kind regards, Samuel Lourenço

[xavier60]

My main point is, it is difficult to make Current Control work properly with a Voltage Follower output stage.

What do you mean? I think there is some confusion. IC7, the op-amp responsible for the CC regulation, is not a voltage follower. The circuit in the OP works, but is not very well tuned.

There are two DACs: IC3 sets the voltage and IC5 sets the current limit. The only voltage follower here is IC4A, and it is needed as the IC3 DAC buffer. IC4B is the error amplifier, IC6 amplifies the sensed voltage in order to "read" the current and IC7 acts as a comparator of sorts that sets the CC mode when the current is exceeded.

Kind regards, Samuel Lourenço

I mean the output stage transistor Q4. I don't fully understand how everything works , mainly what I have discovered from experimenting and somewhat from what others have said.
Having the output transistor functioning as a voltage follower makes voltage regulation easier because the transistor itself is performing voltage regulation in having the Emitter voltage follow the Base voltage. The CV error amp only needs to make small corrections.
Current regulation is easier when the output transistor/s are configured to function as a voltage controlled current source or transconductance amplifier. The CC error amp only needs to make small corrections.
When one has the choice of only one of the output stage types, the transconductance type is the best compromise, from my experience.
It can be made to work well for CV and CC modes.

[bloguetronica]

Today, I've been doing some tests, and concluded that a capacitor in parallel with R5, R7, or in parallel with any resistor that is located in series after the output of a compensated op-amp, is a bad idea. A capacitor from the output to the inverting input is generally a good idea and works well. In my case, when testing modifications to the FAU200 and FAU201 boards, I found out that a type I compensator consisting of a 1K resistor at the inverting input and a 3,3nF capacitor from the output to the former input works well. The noise generated by the DC-DC converter is around 60KHz, and gets reduced.

My main point is, it is difficult to make Current Control work properly with a Voltage Follower output stage.

What do you mean? I think there is some confusion. IC7, the op-amp responsible for the CC regulation, is not a voltage follower. The circuit in the OP works, but is not very well tuned.

There are two DACs: IC3 sets the voltage and IC5 sets the current limit. The only voltage follower here is IC4A, and it is needed as the IC3 DAC buffer. IC4B is the error amplifier, IC6 amplifies the sensed voltage in order to "read" the current and IC7 acts as a comparator of sorts that sets the CC mode when the current is exceeded.

Kind regards, Samuel Lourenço

I mean the output stage transistor Q4. I don't fully understand how everything works , mainly what I have discovered from experimenting and somewhat from what others have said.
Having the output transistor functioning as a voltage follower makes voltage regulation easier because the transistor itself is performing voltage regulation in having the Emitter voltage follow the Base voltage. The CV error amp only needs to make small corrections.
Current regulation is easier when the output transistor/s are configured to function as a voltage controlled current source or transconductance amplifier. The CC error amp only needs to make small corrections.
When one has the choice of only one of the output stage types, the transconductance type is the best compromise, from my experience.
It can be made to work well for CV and CC modes.

I've far more concerned with the CV performance. The CC mode is accessory in this project, and was only added as a safety feature in order to protect the loads.

Kind regards, Samuel Lourenço

[David Hess]

I've far more concerned with the CV performance. The CC mode is accessory in this project, and was only added as a safety feature in order to protect the loads.

This points to a difference between common power supplies which have an adjustable current limit and more specialized power supplies like those found in source-measurement instruments.  Good constant current performance requires low output capacitance and low output capacitance also limits surge current when a lower voltage device is connected to a higher voltage output.

[bloguetronica]

Well David, my objective is to make this supply as versatile as possible, while stable. I'm planning to add just the needed capacitance and a little extra. But CV performance is primordial. It must be low noise and high precision in terms of voltage. The current setting is not nearly as precise.

[Doctorandus_P]

I once tried setting the ouput of my power supply to 30V and the current limit to 20mA, and then connect a led.
It should just light up with 20mA right?
(Note: In those days a led really needed 20mA to be somewhat visible).


In my case the led exploded. Didn't expect that.
But over time you learn.
The power supply was ok

[Kleinstein]

Ideally a LED would just get the set 20 mA. However with real life power supplies there is usually a much higher initial current. This extra current comes from the capacitor at the output, that can be quite large in some cases (e.g. up to 1000 µF for some 3 A supplies) and also from the time it takes for the current regulation to set in.

Here one kind of has the choice of 2 evil. There are mainly 2 types of regulator circuits: those with an output stage that sets a current and those where the output stage sets a voltage with a low impedance.
For the 1st. type a reasonable size capacitor in mandatory for stability, but the cross over to CC mode can be fast. If the regulator is fast and well tuned the capacitor can be relatively small - so no real need for 500 µF, but some 10 µF may be required.

For the 2 nd type one does not need a capacitor at the output or least can use a very small one (e.g. 1 µF). However the onset of CC mode can be slow, especially if the voltage needs to drop. So there is a kind of simulated capacitance, that can be about the size required in the other case. Often there is also some delay before the current limit sets in.  This is sometimes good (e.g. when testing a SMPS or fan) and sometimes bad (e.g. the LED).

So it is really difficult to get a fast acting current limit. Normally one can live with the limitations. There is no need to connect a LED to the supply set to 25 V and hope for the current to be fast enough - set the supply to 0 V and only turn up the voltage after the load is connected.

[bloguetronica]

I can only assess the capacitance that is needed when I have the board on my hands and when I manage to solve the issues. Everything else is a guess. Anyway, it wont be 10uF, as it is too low for this design.

Kind regards, Samuel Lourenço

[Kleinstein]

The circuit shown here uses a low output impedance power stage and could in principle work essentially without capacitor at the output. So 10 µF is not such a bad target for this type of circuit. The difficulty in this circuit is more in getting the current control work well, especially if the differential amplifier is not perfect.

[bloguetronica]

The circuit shown here uses a low output impedance power stage and could in principle work essentially without capacitor at the output. So 10 µF is not such a bad target for this type of circuit. The difficulty in this circuit is more in getting the current control work well, especially if the differential amplifier is not perfect.

Well, today I had an epiphany about that differential amplifier. I think I should try a 10K resistor in series with that capacitor in the feedback from the output to the inverting input. That, in conjunction with the 10K resistor that goes in series to the inverting input, will cause the AC gain to be -1. I think that the AC gain of that stage was set too low, to almost 0, if that makes sense. Perhaps then, it is possible to eliminate that extra capacitor named C12.

Given that all conditions are stable (all op-amps compensated), then it is a matter of testing at full load, in order to find the minimum capacitor.

Kind regards, Samuel Lourenço

[bson]

The IC4B based sense amplifier lacks external compensation.

The compensation is done externally, via shunts in J5.

I haven't had time to follow this thread, but an error amplifier can't have the same or greater bandwidth than the response of that which it controls, be it a voltage regulator or VCO.  The way to control its bandwidth is through external compensation, and that needs to be local to the amplifier.  It can't be shared.  If it has excessive bandwidth it will swing wildly when the downstream stage can't respond.

Edit: oh, and a shunt is a bad compensation mechanism; it works by presenting a load that makes the error amplifier operate current limited.  Since it's shared by the input of the next stage, any change to the input impedance will affect the error amplifier's bandwidth!  And then there's the potential for power dissipation in the error amplifier when it has to drive a large shunt cap.

[Kleinstein]

The IC4B based sense amplifier lacks external compensation.

The compensation is done externally, via shunts in J5.

I haven't had time to follow this thread, but an error amplifier can't have the same or greater bandwidth than the response of that which it controls, be it a voltage regulator or VCO.  The way to control its bandwidth is through external compensation, and that needs to be local to the amplifier.  It can't be shared.  If it has excessive bandwidth it will swing wildly when the downstream stage can't respond.

This limitation is not very strict: if the system to control is well known and well behaved it is allowed to have the controlled system to be slower than the regulator. However this is not the normal case. For a simple (dominant pole) compensation one needs one time constant to be much longer than the others - in theory this could be the system to control.

Still it's likely no a good idea to have an OP without a local compensation - even if only to counteract varying speed of individual units.

[bloguetronica]

The IC4B based sense amplifier lacks external compensation.

The compensation is done externally, via shunts in J5.

I haven't had time to follow this thread, but an error amplifier can't have the same or greater bandwidth than the response of that which it controls, be it a voltage regulator or VCO.  The way to control its bandwidth is through external compensation, and that needs to be local to the amplifier.  It can't be shared.  If it has excessive bandwidth it will swing wildly when the downstream stage can't respond.

That makes sense.

Kind regards, Samuel Lourenço

[Marco]

A switching+linear power supply would have an advantage in getting fast CC with an emitter/source follower, it could use a smaller/faster output transistor.

[bloguetronica]

I officially declare this project to be a failure. Assembled the board, and meanwhile the f*king regulator decided to deliver 24V to the whole 5V bus rail, so I had to replace many of the ICs.

Second try, managed to compensate IC4 successfully using a 10nF capacitor in the feedback loop (a 3,3nF one would do, although marginally). However, when loaded, I was having interferences caused by the DC-DC pre-regulator, causing the output voltage to drift a few 10s of mV. Even soldering a 10uF capacitor in parallel with C16 and replacing R15 with a 22K resistor wouldn't solve the issue. Disconnecting the Vret line to the DC-DC made the issue go away. Meanwhile disconnected the 24V line feeding the regulator and pfff. Again!

So, there are some frailties:
- The DC-DC pre-regulator idea must go away;
- The INA180A1 can surge the 5V line if the supply is disconnected, taking the regulator with it. Then, when the regulator is powered, it fries itself and everything on the 5V line;
- The CC must go away, all of it, despite the fact that it works. A simple SC protection should work.

The good news:
- Apparently, a single 47uF capacitor at the output is enough to stabilize the supply, provided that IC4B is compensated using a 3,3nF to 10nF capacitor;
- The temperature limiting works great.

Learned the lesson. Project ditched. Thread closed!

Kind regards, Samuel Lourenço