Easiest way to buffer

[rthorntn]

Hi,

I have an interface board with 16 outputs and 18 inputs spread across two 26-pin IDC connectors, I have a problem with my outputs (12 on IDC_1 & 4 on IDC_2), the FPGA I use to drive the outputs only has about 1.5mA on a pin and the optocouplers (HCPL-2631) need 5mA.

So would you make up a BOB with two SN74CBT3245C on there and four IDC26 connectors and route the 16 outputs through the FETs and the other pins just pass through?

Thanks.

[Zero999]

Why are you using opto-couplers?

Digital isolators don't need such high drive currents.

https://www.ti.com/lit/ds/symlink/iso7840.pdf
https://www.onsemi.com/download/data-sheet/pdf/ncid9401-d.pdf

[SteveThackery]

I didn’t know there was such a thing. Thanks!

[rthorntn]

Thanks, it's an off-the-shelf CNC board so I'm stuck with the opto-couplers.

[Zero999]

To answer the original question: yes, the SN74CBT3245C will work, but is a bit overkill. How about an ordinary line driver IC?

Some examples:
https://www.mouser.com/datasheet/2/149/74AC245-1006490.pdf
https://www.ti.com/lit/ds/scas513g/scas513g.pdf
https://www.mouser.com/datasheet/2/149/74ACT240-888464.pdf
https://www.ti.com/lit/ds/symlink/74ac16244.pdf

[rthorntn]

Thanks, could you please help a beginner out some more and help me understand why (my friends suggestion) was overkill and how your suggestions differ?

[Buriedcode]

If you're requirement is only a few mA, or say 8mA per channel, and high speed isn't of a concern, then a buffer from the logic series HC/HCT/AC/ACT/VHC/VHCT/LCX will do.  The 74xx541 has a nice pinout - all IO in a line, as does the bidirectional 74xx245.  Where xx is the logic family.

the T in HCT/VHCT/ACT is for TTL thresholds, will only work at 5V, but will accept 3.3V logic on their inputs.  The LCX/LVC families are 3.6V max, but 5V tolerant inputs, but 3.6V is enough to drive optocouplers.

Assuming your IO is one-way  - you're using optcouplers so these channels aren't bidrectional, they're either a fixed input, or a fixed output.

So, theres a huge range of options there - I would go for something like the 74HC541 or 74HC245, since the pnouts are nicer to work with, and they're cheap.  Only 8 channel, but for 16 - just use two!.

[rthorntn]

Thanks, I guess the maximum speed on a step pin could be something like 320K pulses per second:

(200 pulses per rpm x 32 microsteps) x 3000 rpm = 19.2M and then divide that by 60 to get seconds.

Lots of variables, 3000 rpm is more servo and they might not support microsteps, not sure.

I see references to a 400kHz stepping frequency on motion controllers.

[Zero999]

Thanks, could you please help a beginner out some more and help me understand why (my friends suggestion) was overkill and how your suggestions differ?

I didn't check the data sheet for the 74CBT3245 properly.
https://www.ti.com/lit/ds/symlink/sn74cbt3245c.pdf

The 74CBT3245 is completely unsuitable for your application. It simply an array of switches, all controlled by one signal, on pin 19. When pin 19 is low, then each A pin is connected to the respective B pin, via an internal transistor.

Thanks, I guess the maximum speed on a step pin could be something like 320K pulses per second:

(200 pulses per rpm x 32 microsteps) x 3000 rpm = 19.2M and then divide that by 60 to get seconds.

Lots of variables, 3000 rpm is more servo and they might not support microsteps, not sure.

I see references to a 400kHz stepping frequency on motion controllers.

Line driver ICs are several orders of magnitude faster.

[rthorntn]

Thanks, should I just grab the first one on your list, the 74xx541, any idea if there are versions that gave more buffers like 16?

I have a Q over here about being able to switch "around" buffers so I can use this "adapter" on different breakouts:

https://www.eevblog.com/forum/beginners/compact-and-intuitive-way-to-reroute-pins/

[Zero999]

A 16-bit line driver has 16 channels.
https://docs.rs-online.com/5e90/0900766b80024c9a.pdf
https://www.ti.com/lit/ds/symlink/74ac16244.pdf