EEZ Bench Box 3 (BB3)

[jan28]

I setup up the LED steering this way because I couldn't find a pin on the I/O expander to drive it from software.
It should be off if there is no remote programming and on if there is either remote voltage or current steering taking place. This is not wired correctly in the diagram: The 2n7002 gates should be connected to the aux_uset and aux_iset signals, not the output of the AND gates.


I choose the TL072 because it is already widely used in the DCP405, keeping the BOM simple and power rails are available.

You're right protection is not present yet, but should be added in final design. The current rprog on the DCP405 has protection.

The comparator and and gates are there to enable a setting a limit on output when remote programming is enabled.

Two questions I have, while we are discussing this:
- The steering range is from 53mV - 2.5V, not from 0V-2.5V is this a problem, should we add something to compensate for this?
- Instead of limiting the output (in this design), is scaling a better option (more complex to realise). With scaling I mean that if we set a max of 10V on the BB3, 2.5V remote programming voltage gives a 10V output?

[prasimix]

- Instead of limiting the output (in this design), is scaling a better option (more complex to realise). With scaling I mean that if we set a max of 10V on the BB3, 2.5V remote programming voltage gives a 10V output?

Yeah, scaling (by input attenuation) that 10 V gives max voltage makes sense. Many control signals are in 0-10 V range.

[Kleinstein]

Scaling in such a way that 10 V at the input would give the set value may be better, but also much more difficult to implement, as it requires a kind of multiplication.
So I think the way it is drawn is OK: have the external value and the set value as an upper limit.
Instead of the comparator and switching part, one could also use precision clamps, though this would need a negative supply. The switching solution as should be OK, maybe with a small extra filter cap so that the external signal is not too fast.

For the OP I would suggest something like MCP6001, so with a rail to rail output and singel supply (or RR) input.

For the logic polarity one could use NAND gates and swap the signal inputs at the switches. Alternatively one could use 2 diodes as a wired OR instead of the 2N7000 and drive the LED from the AND gates.

[prasimix]

OK, take into account that we have on disposal -5 V and about -8 V if negative supply is needed.

[jeremy]

Hi everyone,

I have noticed that when I hook up my multimeter directly to the output of the DCM220, it reads ~1.3V when the outputs are off or inhibited. This is not the case for the DCP405 modules; they read as basically zero.

Is this behaviour expected? I realise that the load of the multimeter is very small, but I just wanted to check if anyone else sees this. I had assumed that OFF = 0V.

Thanks,
Jeremy

[prasimix]

Yes, that is expected it hasn't down programmer as DCP module, and it cannot go below 1 V even with load connected. Please note that DCM is "auxiliary" DC source and cannot be compared to the capabilities of the DCP module. What you're measuring is remaining voltage on output capacitor and if you try to measure current it will be close to zero.

[jeremy]

Thanks for the clarification, and yes, I am aware it is more for powering less sensitive/auxiliary systems. Are there current limits applied when the module is in the off state? If I was to short the + and - while the supply is off, would that damage the unit?

[prasimix]

No, current limit (that is very coarse in comparison with DCP module) is not active when LT3763 is off, but only remaining current from output capacitor will flow and that shouldn't have any effect on module itself.

[prasimix]

This is a new proposal for Remote programming presented in #498.
Input op amp is changed to MCP6001 powered from single rail. LED driver is also changed, and 2G08 instead of 2 x 1G08 is used. I also added 10K on all comparator inputs.

[prasimix]

I forgot to mention that LM293 is now also powered from single rail that output stays within 0 to V+ range.

[Kleinstein]

I don't think the 10 K resistors at the comparator are needed. The signals are well limited in range.

For some filtering it may help to have capacitors in parallel to R11 and R4.

I don't think one would need Q1. Just the LED towards ground should be OK.

[prasimix]

Thanks @Kleinstein, corrections below:

[jan28]

I think @kleinstein mentioned it before, in this design the LED is ON only if vrprog or iprog is on AND the input signal is below the limit. This means that if you turn on rprog and you input voltage is below the limit, the Led will turn on (just as it works inter current dcp405). If you increase your rprog input voltage above the limit, the led will turn off although rprog is still set.

If the anodes of D1 and D2 are connected to aux_uset_sw and aux_iset_sw then the behaviour of the leds changes: they are always on if prog is on, even if the output is limited by the set limit.

In my opinion both are valid, just different:
- With the current firmware there also is an indication rprog is on in the display in the set values. You will also be able to see on the main screen if it is voltage or current programming. The led has some redundancy to this.
- It can be a nice indication you’re rprog input is getting limited because there is no other way to know if limit is enforced or not. It might confuse users because the behaviour requires reading the manual....

A third (fancy, feature creep) option: make the led dual color. One color (yellow?) for rprog active and another colour (red?) if rprog is active AND the limit is enforced.

More feature creep (or an enabler for later software features): feed the led status as it is now (or the OR of the two comparators) back to the MCU. This allows for measuring if limit was hit and maybe optionally enable a protection on it in software (e.g. turn output off if limit is hit) in the future.

One question I still have: Is the limit circuit (comparator, logic and relais) in combination with C21 dampening and the limited bandwidth of the output (500Hz ~= -3db) fast enough to not overshoot beyond the limit of you step up rprog above the limit? This circuit will probably pass a (small enough?) pulse above the set limit into into the next stages of the dcp405, these stages have a limited bandwidth that probably smooths the step enough to not have any overshoot, but it should be verified. If I have time I will try to measure the max pulse width not causing overshoot on he current rprog interface (voltage only...) that will give an estimate of the budget for the delay of the comparator, and gate and relais combination.

[Kleinstein]

I see the main purpose of the LED to show if the voltage or current is actually set by the external value. So the LED turning off indicates that the setting is by the internal setting - either because of hitting a limit or because programmed to do so, which is the normal case.

The speed of the external signal causing overshoot is a valid point. The idea of the C21, C6 is to slow down the external signal enough so that the comparator / switch is fast enough to avoid much overshoot. It won't be a 100% protection as there is no good control how hard the input is driven extenally. The comparator is quite fast giving a dealy of maybe a 2 µs or so (depends on how much the limit is exceesed) the logic CMOS swtiches are very fast (e.g. < 100 ns). The trasition from the external to internal setting is the faster one - depending on the values of R2,R3 the transition back  can be a bis slower, which may be good. So I would consider somewhat larger values for R2 and R3.

I just realized that the clamping diode for the input should be more like 1 pair of back to back diodes instead of a zener / transsorb. The voltage accross the inputs should normally be small anyway, and there are diodes to the supply inside the OP anyway.

[jan28]

I did some tests to simulate the possible overshoot using a DCP405 and driving the prog of it with a signal generator and an on-amp configuration like the design.

The AWG starts at 0V then generating a short pulse at 2.5V (full scale of rprog, rise time = 9ns) and after that it goes down to a lower value simulating the situation where the rprog is driven at a step of 2.5V and there is a limit set.
I varied the voltage levels, the length of the pulse and the load of the DCP405:
- The load doen't change things as long as the current limit is high enough.
- The worst case seems to be a low limit value and a full scale (2.5V) step. Seems logical..
- Worst case meaning the settings where the shortest pulse width created overshoot.

First I played around with limit around 1.25V (20V output). A pulse < 100us didn't give overshoot.
Playing with low limit values (around 1V) make things a lot worse:

This is with a pulse of 20us at 2.5V and a limit voltage of 100mV (output = 0,7V)

Decreasing the pulse with to 10us gives:


Going even lower (limit = 70mV / output = 0,2V) also gives overshoot at 10us, reducing the pulse with 8us eliminates this also.
This is a pretty extreme edge case: You set the output limit to 0,2V and drive the rprog interface to deliver 40V...

The most extreme overhoot (relative) is to step 8mV into rprog with an offset of 57mV (= 0V output in my setup), still feeding a 2us 2.5 pulse, then  8mV causes a output of overshoot of 100mV at 120mV. It is a very extreme case: who will set the limit to 0.1 V?

Given the specs from the datasheets of the comparator and logic this should not be a problem. The comparator is max 2us and the logic is an order of magnitude below that.

I think the limit circuit is fast enough for the voltage control.

[prasimix]

@Jan28, what current range settings you're using? If you are in "Best" mode then you can expect huge overshoot since firmware based current range switching is too slow.

[prasimix]

I think @kleinstein mentioned it before, in this design the LED is ON only if vrprog or iprog is on AND the input signal is below the limit. This means that if you turn on rprog and you input voltage is below the limit, the Led will turn on (just as it works inter current dcp405). If you increase your rprog input voltage above the limit, the led will turn off although rprog is still set.

If the anodes of D1 and D2 are connected to aux_uset_sw and aux_iset_sw then the behaviour of the leds changes: they are always on if prog is on, even if the output is limited by the set limit.

Yes, I understand that and as @kleinstein explained I don't see any issue with it at all.

In my opinion both are valid, just different:
- With the current firmware there also is an indication rprog is on in the display in the set values. You will also be able to see on the main screen if it is voltage or current programming. The led has some redundancy to this.
- It can be a nice indication you’re rprog input is getting limited because there is no other way to know if limit is enforced or not. It might confuse users because the behaviour requires reading the manual....

On the firmware side, we have enough freedom to present this special situation in the most convenient way and that we do not need to consult the manual for that .

A third (fancy, feature creep) option: make the led dual color. One color (yellow?) for rprog active and another colour (red?) if rprog is active AND the limit is enforced.

More feature creep (or an enabler for later software features): feed the led status as it is now (or the OR of the two comparators) back to the MCU. This allows for measuring if limit was hit and maybe optionally enable a protection on it in software (e.g. turn output off if limit is hit) in the future.

Um, you need to consider that this is already a dual LED THT and I don't know if there is a version where one of them is also bicolor.

[prasimix]

The speed of the external signal causing overshoot is a valid point. The idea of the C21, C6 is to slow down the external signal enough so that the comparator / switch is fast enough to avoid much overshoot. It won't be a 100% protection as there is no good control how hard the input is driven extenally. The comparator is quite fast giving a dealy of maybe a 2 µs or so (depends on how much the limit is exceesed) the logic CMOS swtiches are very fast (e.g. < 100 ns). The trasition from the external to internal setting is the faster one - depending on the values of R2,R3 the transition back  can be a bis slower, which may be good. So I would consider somewhat larger values for R2 and R3.

10K instead of 4K7 for R2, R3 or even higher?

I just realized that the clamping diode for the input should be more like 1 pair of back to back diodes instead of a zener / transsorb. The voltage accross the inputs should normally be small anyway, and there are diodes to the supply inside the OP anyway.

I don't understand this, the input voltage can be 0 to 2.5 V, and there was talk of a 0 to 10 V range. How to protect it only with a pair of back to back diodes?

[jan28]

@Jan28, what current range settings you're using? If you are in "Best" mode then you can expect huge overshoot since firmware based current range switching is too slow.

It was in "best mode" with the limit set to 5A. So I think there was no software switching?
Most of the measurement I did with 10ohm resistor.

[prasimix]

Just to be sure, check how it looks like with current range set to 5 A.

[jan28]

Checked with 5A setting: No difference, looks the same as previous screendumps.

[Kleinstein]

Most of the protection would be from the MCP6001 internal diodes to the supply. The diodes between the inputs are more like a fine print to that and may help a bit with transients.
The MCP6001 is specified with up to 2 mA at the input (which is rather low for an OP). With 10 K at the input this would be good for some 20 V.  Because of the low values one may consider raising the resistors a little (e.g. 27 K), e.g. to allow the 50 V from the supply itself not to damage in input. Alternatively there may be a more tolerant OP (AFAIK TI has a similar one).

[jan28]

I forgot to mention that LM293 is now also powered from single rail that output stays within 0 to V+ range.

For limiting the output of the opamp this works, however:
This limits the common mode input range (or how much it can ""float"") of rprog with respect to the negative power output plug of the DCP405, doesn't it?

The whole DCP405 module is floating with respect to protective earth. In the DCP405 the negative input plug is a little above the GND of the module (mainly due to shunt). The opamp gives the DCP405 a differential input for rprog (X1-1 and X1-2 where X1-2 is not connected to GND of the module anymore). This differential input can be operated between the range of the power of the opamp it drives meaning the negative rprog input should be between 0 and  +2.5V relative to negative power output plug. This is enough to allow connecting the rprog- and the negative output to the same potential (e.q. protective earth). I think this is an important use case (at least for me).

If a opamp with boarder input power is used the range also gets increased. E.q. it is also possible to have rprog+ at the same potential as the negative output.

I hope I'm not creating confusion.   

[prasimix]

I hope I'm not creating confusion.   

No, you didn't, after all you started this interesting enhancement, and I'd like to add it into the next revision.
If LM293 is powered from dual rail then we need to limit "low" level with diode or something otherwise -V will appear on its output that I believe is not something that 2G08 would like to see on its input.

[prasimix]

Most of the protection would be from the MCP6001 internal diodes to the supply. The diodes between the inputs are more like a fine print to that and may help a bit with transients.
The MCP6001 is specified with up to 2 mA at the input (which is rather low for an OP). With 10 K at the input this would be good for some 20 V.  Because of the low values one may consider raising the resistors a little (e.g. 27 K), e.g. to allow the 50 V from the supply itself not to damage in input. Alternatively there may be a more tolerant OP (AFAIK TI has a similar one).

Huh, you're right, I overlooked that MCP inputs are floating. As @jan28 explained question is do we need to power it with dual rail, and in that case we have to search for other candidate since it cannot accept +/-5 V what is currently available.