Closed gkasprow closed 3 years ago
- Do we need PoE?
Not strictly need it if causes problems. But it fits the power envelope and would play nicely with the rest of the hardware.
- I assume that we can connect all AFEs to the same SPI interface, 4 chips per SPI
Yes.
- DACs will share same SPI interfaces - two chips per SPI
Yes.
- do we need individual PWMs for each TEC channel?
You mean one MAX1968 per TEC? Yes. Or do you mean the I/V limit PWMs? Those can be banked, let's say two TECs per set of voltages. But let's try to keep them as separate as possible. But I might be misunderstanding you.
- I assume we want a downstream/upsteream EEM channel like in the Stabilizer. Was it tested already?
With Stabilizer not that I know of. With Humpback and Urukul yes. If there are space/pin problems, I'd rather drop the EEM connector but keep PoE.
- what connectors? We can use ribbons with DSUB placed on separate panels as in the original Thermostat. But this takes a lot of space. We can also use HD DSUB and have 2 TECs and 4 sensors in each HDSUB15. In this way, noone will mix them with Thermostat or RS232 :) These days nobody is using VGA.
I'm only waiting for someone to connect a monitor to a Booster main board ;)
Do you mean DE-15? Fine with me. @j-hirsch and FBH should weigh in.
I tried but in case of DE-15 we don't have enough pins. We won't fit two DA-26 Let's use one DB-44 (High Density D-SUB) which has enough pins and there are ready to use extension cables available. I will design a breakout board to make testing easier
The pinout could be as follows
It looks like we will have individual PWM channels for all TEC driver settings.
Initial placement
When running one big tec from two channels in parallel, would we connect them on the header board?
I'm not 100% sure if you can easily run two channels in parallel. These are synchronous converters - even a small differences in duty cycle will result in high cross-current flowing within the pair of shorted outputs. The header board is for debugging purposes.
Aren't the drivers effectively current sources which should work in parallel? FBH indicated that they need one case with two in parallel. Best to disuss with @j-hirsch @schiemangk.
Alternatively it might be easy to just use two electrically separate half-size TECs to get the same result. Having two smaller independent TECs zones may actually be much better.
looks nice! Any further thoughts about power/thermal management here? Can you get enough juice out of a single EEM for that many TEC drivers?
It'll need active power management and load sharing anyway. EEM is probably not going to be the use case for us but the connector/cable is probably not going to be the limitation. It's just there because we can.
@jordens