AndyUK's Build - 1.2x1.0m Gantry

[Kitwn]

You learn something new every day!

A scan of the RS catalogue shows you can get inexpensive devices using Darlington output stages with CTR values as high as 50 and more if you have the luxury of designing the circuit yourself, though they have a higher sturated output voltage. Problem is we're often dealing with couplers already installed in larger devices and cannot choose.

Pull-up/down resistors are often already included in many devices, it being a very bad idea to leave logic inputs to float as the mood takes them. This is how the common "enabled if left unconnected" principle is implemented.

[AndyUK]

Andy,

I think you understand a fair bit on the old electronic side. I was casually browsing this thread and one thing is niggling me. Opto isolators typically have a low CTR - usually significantly less than 1, often less than 0.5. That does mean that cascading optos is going to end (eventually, depending on CTR and number of optos in chain) with the latter optos not switching effectively. Really, the opto should be switching a high-z logic device rather than the relatively high current LED input on the successive circuit.

You link a rating of the output at 100mA - that /could/ just be the Ice(max) for the PT in the opto, rather than an actual expected switching current (reading the 4n25 data sheet this is the case).

I'm not suggesting this is your problem, but bear it in mind and measure the switching voltages along the opto-cascade,

Hi Doddy,

Thanks - that is quite high praise, but I'm still very much stumbling my way around so don't assume too much!

I hadn't really considered the CTRs of the optoisolators, however is this relevant if I'm effectively connecting each drive in parallel with oneanother? My intuition is only really concerned about the fact that I have the Optoisolator from the drive, and then the one in the OSSD output which as I understand is effectively acting as a transistor to ground. Upstream of the drive is merely the current limiting resistors and the +24V rail. My ~40mA came from the sum of 10mA per drive based on the current limiting resistors, and if the CTR<1 then I shouldn't have damaged anything staying under 70mA.

I suspect I've totally missed the point here though :) Please educate me further!

[Doddy]

I suspect I've totally missed the point here though :) Please educate me further!

I'm trying not to mis-direct this thread (tangent?, what tangent...), but to explain my logic....

If I redraw your schematic (lesson #1, "Eagle" is not good for this)



and zoom into a bit of text that I torturously added



If you /assume/ that the opto-isolator on the UB1 OSSD output has a forward current through the LED (I1) of 10mA (not unreasonable, and likely a design goal). Also, /assume/ the CTR is 0.5 for that device. That means that the switching current (Iossd) would be I1 * CTR = 5mA.

Now, Kirchoff's law says basically that the sum of the currents at a node equals zero. So, if 5mA is flowing into the OSSD input, then the sum of currents through the 4 separate '806 drivers must also equal 5mA. Give them the benefit of the doubt and say that all '806's are created equal, then your forward current through each '806 LED is that 5mA, divided equally by 4, or 1.25mA. Not the 10mA that you believe you have. Then you're at the mercy of the circuit design in the '806 behind the internal opto-isolator. That 1.25mA is multiplied by the local CTR and further diminished. Also, the CTR is impacted by the forward current - it's less at 1.25mA than if driven by 10mA, and so the internal switched voltage swing is further diminished, and may, or may not impact on the performance of the ENA input.

The bit of maths around Vossd is easily verified by yourself - measure the voltage, wrt ground, at the OSSD output when you have a fault condition. If the OSSD output was an ideal switch you'd expect it to be 0V. I expect it to be rather higher (calculated around 20V), based on the less-than-ideal switching behaviour of the opto-isolator (and if I have to correct my auto-correct for opto/onto once more....).

I'm cautious of playing too much behind a calculator - measure the voltage yourself. Reality beats theory any day. The one thing that worries me about my theory is that I would expect the behaviour to be accurately described in the owner's manual of the UB1. And the manual doesn't describe my perceived behaviour - so I could be wrong here.

I've tried to look online for a high-res image of the UB1 for information on the opto-isolator chosen - to no avail. If you can check and advise? It could be an intelligent selection of a device with an unusually high CTR. Pretty sure they are not buffered devices as they look to be 4-pin devices.


EDIT:

Here's an idea. Short the OSSD output to ground (or just the wire from the OSSD output). Either that will work as expected (inhibit each driver) or not. That's a quick way to fault-find the circuit.

[Lee Roberts]

If I redraw your schematic (lesson #1, "Eagle" is not good for this)

Sorry to side track Andy...

Doddy have a look at Fusion 360, they've been working on updates for the PCB design side more recently, I don't know if schematic support is available as I haven't had time to explore just yet...Autodesk owning Eagle i'm now thinking probably not.

[AndyUK]

Managed to reprogram one of the drives last night (its a faff now they're in the cab) to be Enable Active low, and instantly was able to get my first self-powered movements on that axis (huzzah!! - I'll post a video once I have all three axis working and this enable issue resolved, because as usual the garage is a tip).

I didn't change the enable wiring, and noticed that the drive was enabled whatever happened (one of the other unconnected drives faulted out which dropped the OSSD output, but it had no effect on the active low drive). Therefore, the problem isn't that I've got my logic states messed up, or at least that's not exclusively the problem.

My next step is to try experimenting with wiring setups to try and get the enable functioning. First I'm going to try Vsarak's two suggestions above, first utilising the 5V rail and another output (schematic B), and second, using a differential connection from a spare axis output (schematic A).

The third step would be to try the NPN connection recommendation in the EM806 manual (which I now realise is effectively what I've been trying to do). The only difference here is that the resistors are placed on the ENABLE - line rather than the ENABLE + line - would that really make a big difference? See attached.

[AndyUK]

Wow thanks for that Doddy, I see what you mean now :)


What I think are the optoisolators (4pin IC directly above each output) have 838 written in small font above a larger central "P185". Bottom row is "GB JG".

Also, how the hell did that board get so dusty?

Measured the voltage but wrt 24V not ground (oops). Withe all drives on and connected, OSSD output off (i.e. fault condition drive should be disabled) I get 0V reading (so the OSSD pin is at 24V). With the OSSD output on (drives should enable) I get a reading of 23.3V (so OSSD pin is at 0.7V).

Connected a single drive to 5V and the OSSD pin without resistors, worked perfectly. Enabled and disabled the drive like a champ exactly as expected in both active low enable mode and active high enable mode (enabled in opposite situations obviously). Will try connecting all drives simultaneously next, and I guess worst case will just run them through a spare relay operated by the OSSD pin?

[Doddy]

Likely a TLP185 or equivalent. CTR = 100-400%. Your measured voltage is fine and indicates the behaviour should be as you expected. Disregard all my random ramblings about optos.

[Kitwn]

Likely a TLP185 or equivalent. CTR = 100-400%. Your measured voltage is fine and indicates the behaviour should be as you expected. Disregard all my random ramblings about optos.

As someone who likes to design and build the odd circuit of his own, I have found your 'ramblings' extremely informative.

[AndyUK]

I have found your 'ramblings' extremely informative.

Seconded, this train of thought has more than doubled my knowledge of Optos.

[Doddy]

Lee, the work auto desk have done is to integrate the eagle board design together with 3d models of components to post forward into F360 for correlation between electrical and mechanical design. Think of it as a 3D viewer for Eagle, but with obvious advantage for designing the mechanicals around the electrical design.

It’s neat when it works but a complete buggeration from a component library manager POV.

Unfortunately it doesn’t make schematic capture any better (though in fairness autodesk have developed Eagle more in the last couple of years than any of the previous owners)

Sorry!, Andy, thread hijack