3.3V voltage repeater

Status
Not open for further replies.
S

someteen

Well-known member
I need to read some Hall effect current sensors (Allegro ACS758 series) thus I have to feed them with the same reference voltage as Teensy 3.1 ADCs. As I'm using a 2x14 pins ZIF socket for Teensy 3.1 module, I can't access the bottom side AREF pin hence I couldn't use an external reference for both Teensy and current sensors.

The only(?) solution is to use Teensy 3.3V output (used for internal ADC reference) to feed the current sensors. Anyway, I'm going to use at least 8 current sensors, each needing a max. 13.5mA supply current thus exceding the 100mA datasheet limit of that 3.3V output.

What's the best way to build a (high accuracy) voltage repeater for that 3.3V output, with enhanced current capabilities?
 
I'd suggest a external power supply with a very high PSSR to provide the power to the Allegro chips (ADP150?).

They should share a common ground with the Teensy. If you want the ADP150 to be your external reference as well, simply remove the 470 Ohm resistor topside on the teensy and use a POGO pin to make contact with AREF from below.

I would not call the allegro series to be high precision, however. IIRC, the manufacturer claims something like 5% accuracy for their series. It's one of the reasons I abandoned their chips for my power measurement project.
 
Last edited:
Thanks a lot for your suggestions! But, like I've said in the prior post, I can't access Teensy AREF pin (the ZIF socket doesn't allow that). Or maybe I could solder a pin on the upper side then run a wire to the external reference?

Regarding current sensor accuracy, the datasheet mention 1.2% (full range). Could you recommend a better alternative? (I need it to be bidirectional and having galvanic isolation)
 
someteen said:
Thanks a lot for your suggestions! But, like I've said in the prior post, I can't access Teensy AREF pin (the ZIF socket doesn't allow that). Or maybe I could solder a pin on the upper side then run a wire to the external reference?
Click to expand...

If you have to use a ZIF socket, I suppose the wire is the best way to ensure that the AREF voltage = the supply to the ACS chips.

someteen said:
Regarding current sensor accuracy, the datasheet mention 1.2% (full range). Could you recommend a better alternative? (I need it to be bidirectional and having galvanic isolation)
Click to expand...

I went with the MCP3910 series of energy measurement chips because the Analog Devices stuff was nigh impossible to calibrate as suggested by the OEM. I simply do not have the equipment here to produce consistent 115VAC signals with a 0.5PF, for example. The MCP3910 is relatively easy to use with the Teensy (runs on 3.3V), offers a PGA, etc.

The measurement side is a "revenue-grade" CR magnetics current transducer and a small voltage transformer. I happen to think this is a better solution, esp if you're dealing with a high-current situation. Perhaps I'm overly cautious but I have little faith in 1oz copper traces carrying 16A of current, no matter how well you stitch them, fill vias with solder, use externally applied solder as an additional conducting layer, and so on.

However, there never is a free lunch. The use of a transformer and a current transducer ensures that the two signals will be out of sync and will have to be re-assembled properly after measuring them (I measured a 1.3ms lag between the transducer and the transformer). But this approach minimized the number of parts on the high side that a later UL certification will be subject to.
 
Last edited:
@onehorse:
That's a nice Teensy shield, too bad I can't wait that much.

@Constantin:
I have to monitor currents/voltages for a photovoltaic system (power inverters, battery bank). I was considering some 200A/100A/50A bidirectional current sensors (ACS758) as I have used them before with good results.

Actually, I don't think I need such a high accuracy (some sensors just read the currents through different branches of a lead acid battery bank (series/parallel) and I'm more interested if those currents are well ballanced).

Those MCP3910 sensors are certainly better but I think I'll choose the simplicity and availability (I already have some spare) of Allegro current sensors.

Back to my problem, it just crossed my mind a possible solution (for a voltage repeater).
What if I'll generate a PWM signal to drive a simple buck regulator (5V > 3.3V) using one of the Teensy outputs then read the converter 3.3V output using an ADC port (same Teensy). This way, I'll manage to get a negative feedback network, by further changing the PWM signal as to maintain that generated 3.3V voltage at a steady value (identical to ADC 3.3V reference).
 
Actually, I have a better idea: I'll use a LM317 regulator to generate a 3.3V reference for the current sensors. Next, I'll read this voltage using an ADC port then I'll compute the ratio between Teensy internal reference (unity) and this external (sensors) reference. This ratio will be used to adjust the sensors voltage readings.
 
someteen said:
I'll use a LM317 regulator to generate a 3.3V reference for the current sensors. Next, I'll read this voltage using an ADC port then I'll compute the ratio...
Click to expand...

You might want to choose the LM317's resistors for slightly less than 3.3V. The ADC is only able to read up to whatever its VCC voltage, which is nominally 3.3V. If the LM317 voltage happens to be slightly higher than the Teensy's VCC voltage, you won't be able to accurately read it. Just a slight change in the resistors can alter the LM317 so its voltage is somewhat less, and always readable so you can use it for calculating the ratio.
 
Thanks, Paul! That's what I thought (but forgot to mention) about it, too. I'm going to set the LM317 to 3.2V or something (to remain subunity) though I have some doubts about thermal stability/accuracy of both Teensy internal reference and LM317 output. I might set an alarm when the external 3.2V reference becomes (over)unity.
 
Status
Not open for further replies.
Back
Top