help please with toroidal transformers selection

[Neale]

I don't want to get hung up on this transformer voltage business (except that I was surprised by the toroidal transformer I bought recently that measured about 5% above nominal voltage when off-load - the "regulation" factor that's mentioned above which I hadn't expected) but I am interested in the practical choice of smoothing capacitors. I am assuming here that the linear PSU is feeding only the stepper drivers and nothing else, which I think is generally the case.

Stepper drivers can cope with fairly wide voltage ranges. From memory, the EM806 which I am using is rated at 24-80V absolute min/max, for example, although the Leadshine recommendation is to stay at least 10% away from either extreme. We are not building audio amplifiers here, so we're not interested in hum levels on the output. The capacitors are being charged at 100Hz (assuming the usual bridge rectifier) but the steppers are pulsed at a few KHz, so any given pulse will see only the variation in voltage during that pulse - small compared with the variation across a half-mains cycle? On top of all that, one of the stepper driver's main jobs is to control and limit current through the stepper coils, so one thing it can do very well is cope with a range of input voltages while trying to maintain the target current through the stepper coils. So do I really care if the DC output voltage sags by a few volts, across a half mains cycle? In practice, then, just what size capacitor will do an adequate job without going OTT? What is the effect of a too-small capacitor? Don't know the answer myself, but once my machine is back running I'll stick an oscilloscope on the PSU output and see what's happening, just in the interests of science.

[Clive S]

Very well put Neale some sense is prevailing at last.

[Tom J]

In practice, then, just what size capacitor will do an adequate job without going OTT? What is the effect of a too-small capacitor? Don't know the answer myself, but once my machine is back running I'll stick an oscilloscope on the PSU output and see what's happening, just in the interests of science.

1000uF for 1A, voltage is important (20% more than operating voltage and you are safe, min 100V caps for 80V system.)

Small capacitor won't take enough power from stepper when slow down or change direction (EMF). I am scared to write anything now

[m_c]

1000uF for 1A, voltage is important (20% more than operating voltage and you are safe, min 100V caps for 80V system.)

Small capacitor won't take enough power from stepper when slow down or change direction (EMF). I am scared to write anything now

The thing is, that 1000uF per A is an arbitrary figure provided as a guideline.
I just had a quick scan through the Gecko Drive guide, and they list the following formula for minimum capacitance in uF-
(80'000 * I)/V

I'd imagine they've done some calculations and research to reach that figure, and it'll involve the maximum voltage ripple their drive internal capacitors can handle safely under various conditions.

[Clive S]

Anyone wanting to make one Jonathan is selling these http://www.mycncuk.com/threads/10475...al-transformer. A bargain I think.

[Neale]

A few months ago I said that I would have a look at the output of my power supply with an oscilloscope to get a feel for what was happening to the supply voltage with the router in operation. I was able to do this recently. I had some gcode running that was winding X and Y backwards and forwards at full rapid speed (gantry probably weighs in at around 50Kg, rapid speed on X and Y 4.5m/min, NEMA23/ballscrews, to give an idea of mechanical load), and I hung the oscilloscope off the supply to one of the drivers. Basically, there was the best part of 10V mains frequency ripple, with maybe a couple of volts of high-frequency noise on top of that. Mains ripple was presumably what the smoothing capacitors could not remove, and the high-frequency noise was the combined result of four unsynchronised switching drivers. That's all on top of a nominal 68V DC supply - the output voltage was swinging either side of that. The amount of ripple didn't change very much when the steppers were stationary although the stepper drivers are set to half-current mode. Clearly that doesn't make much difference to the PSU ripple.

Edit - have now seen that PSU has 2x6800uF smoothing capacitors, 100V/105C rated. Voltage limit is fine for this job and it's good to see the higher temp capacitors being used rather than the cheaper 85C - electrolytic capacitors are not the most reliable of electronic components but the higher-temp versions are better.

On the face of it, that doesn't sound like a great result. I was surprised by the amount of ripple - much more than expected. I can't read the capacitor values where they are, unfortunately, so I'm not sure how generous they are (the original PSU is the PS806 from Zapp although with a replacement transformer). On the plus side, the stepper drivers clearly don't give a damn and just keep on doing what they are supposed to be doing, and do it very well. Maybe, just maybe, it's possible to get a bit too hung up on the details sometimes and worry unnecessarily.

[m_c]

I can't remember the specifics, but ultimately capacitor sizing comes down to load, and how quickly it charges. I'm hoping somebody more knowledgeable will be along with the more exact details, but I'll give a basic view.

As you know the capacitor has to provide energy during the period between charging cycles. The more load for any given capacitance, the more the capacitor voltage will drop between cycles.
Now if you choose too low a capacitance, you get a high voltage ripple on the output.
Most stepper drives will probably handle this, however by having the capacitor voltage fluctuating too much, you risk overheating the capacitor. As with nearly everything, nothing is 100% efficient, so every time you charge/discharge a capacitor, energy is lost to heat.
This is why large electrolytic caps normally have a vent, so if they do heat up and pressure builds up inside them, they can vent in a safe manner rather than exploding. For completeness, if you look at the end of small electrolytic capacitors, you'll see a scored line, which is a built in weak point, for the same reason.
Another thing is during motor deceleration, a smaller capacitor will not absorb as much energy, so over-voltage becomes a risk.

In terms of a powered circuit, other than cost, there's not really such a thing as too much capacitance, but you will reach a point where extra capacitance won't provide any tangible benefit.
With a large capacitance, you have the issue of start up surges, as a fully discharged capacitor will appear as a dead short at initial power up gradually increasing in resistance until fully charged. The more capacitance, the longer the initial charge will take during power on.
Then you need have to deal with discharging the capacitor after power has been removed. As Jonathan mentioned somewhere already, you could end up with quite a lot of heat being generated by a discharge resistor in order to discharge the capacitor(s) in a reasonable time frame once power is removed.

There are calculations that will work all this out, but it's not something I've done for a while. However I'm sure a google search for capacitor ripple voltage will turn up some detailed results.

[Jonathan]

I don't want to get hung up on this transformer voltage business (except that I was surprised by the toroidal transformer I bought recently that measured about 5% above nominal voltage when off-load - the "regulation" factor that's mentioned above which I hadn't expected) but I am interested in the practical choice of smoothing capacitors.

Another thing to play with is adding or subtracting turns from the transformer to tweak the voltage. But I wouldn't recommend that by design, only if you the insulation materials and enameled wire lying around... It's a neat way to make an auxiliary supply though - just add on an extra winding.

Stepper drivers can cope with fairly wide voltage ranges. From memory, the EM806 which I am using is rated at 24-80V absolute min/max, for example, although the Leadshine recommendation is to stay at least 10% away from either extreme.

Having scoped the turn-off overshoot voltage of the MOSFETs in a DQ860MA, I have no qualms about running them at 80V. They use 100V mosfets and the overshoot was only a few volts, so still some headroom. They do skimp on the high frequency decoupling, so I added an extra ceramic capacitor on the empty footprint. I can't comment on other drives though.

We are not building audio amplifiers here, so we're not interested in hum levels on the output.

Yes!

The capacitors are being charged at 100Hz (assuming the usual bridge rectifier) but the steppers are pulsed at a few KHz, so any given pulse will see only the variation in voltage during that pulse - small compared with the variation across a half-mains cycle?

Correct. The stepper driver is a current controller with a switching frequency of 10-20kHz, so the current control bandwidth will be plenty sufficient to deal with some input voltage ripple.

On top of all that, one of the stepper driver's main jobs is to control and limit current through the stepper coils, so one thing it can do very well is cope with a range of input voltages while trying to maintain the target current through the stepper coils. So do I really care if the DC output voltage sags by a few volts, across a half mains cycle?

Probably not, for a few volts...

In practice, then, just what size capacitor will do an adequate job without going OTT? What is the effect of a too-small capacitor?

That is the question everyone should be asking - the answer is not straightforward.

Don't know the answer myself, but once my machine is back running I'll stick an oscilloscope on the PSU output and see what's happening, just in the interests of science.

Maybe I'll try it this evening, if I can find that spare stepper driver...

Meanwhile, I just dug out these graphs I captured whilst repairing a stepper driver, showing the voltage across one mosfet Vds (blue) and the line current through the same (yellow).

The first graph shows a normal situation - we see the current rising when the voltage is applied, and falling otherwise with the inductance of the motor smoothing this current. Observe how little overshoot there is on the voltage.



For this one I removed the dc-bus capacitor (can't remember if it was the small or large one) and supplied from a lab PSU. You can see the voltage and current now oscillating horribly.



Same again, but with a lower supply voltage - notice the current is the same (so the regulation is working) and the duty cycle has increased.



This doesn't tell us anything about sizing the supply capacitor, still it's mildly interesting...

The thing is, that 1000uF per A is an arbitrary figure provided as a guideline.
I just had a quick scan through the Gecko Drive guide, and they list the following formula for minimum capacitance in uF-
(80'000 * I)/V

This is what I like to call a "Forum Formula": formula's commonly distributed with little concern for how they were derived. Often empirical, commonly including a seemingly arbitrary constant designed to make the user think they are learning something.
Just like the formula's in the first post to estimate the supply current - these bare little relevance to reality.

Where does the 80,000 come from? Which is the I, motor or supply current? etc...

The 1000uF per amp is derived from .. The gecko formula will be something similar.

There is some better information in this thread. Post #2 is pertinent.
This one too.

I'd imagine they've done some calculations and research to reach that figure, and it'll involve the maximum voltage ripple their drive internal capacitors can handle safely under various conditions.

But do they apply to every driver - maybe close enough, maybe not?

Anyone wanting to make one Jonathan is selling these http://www.mycncuk.com/threads/10475...al-transformer. A bargain I think.

Thanks Clive - I still have two left and more capacitors on the way.