Shop OBEX P1 Docs P2 Docs Learn Events
Same threshold voltage for all pins on a given propeller chip? — Parallax Forums

Same threshold voltage for all pins on a given propeller chip?

lonesocklonesock Posts: 917
edited 2012-07-13 11:26 in Propeller 1
Hi, everyone.

I have a little application where it would be very convenient if any 2 pins on the prop had the exact same input voltage threshold (i.e. within a few uV)...does anyone know if this is the case? I couldn't find any info on that from the datasheet. Any ideas on a test I could use to find out, without an external variable voltage source?

thanks!
Jonathan

Comments

  • TubularTubular Posts: 4,706
    edited 2012-07-11 16:34
    Easiest way is to hook up the sigma delta circuit with the 2 external caps - that way you can simply measure the input pin with a (high impedance) DVM. Threshold is around 1.42v for a 3.3v supply. Beau had a single pin sigma delta arrangement iirc.

    I suspect doping density variations and temperature would affect it, so neighboring pins are probably the best bet.
  • Cluso99Cluso99 Posts: 18,069
    edited 2012-07-11 16:54
    My guess is that all pins would be identical, but chips themselves could vary because of their original location on the wafer.
    Perhaps we might hear from Chip?
  • Mike GreenMike Green Posts: 23,101
    edited 2012-07-11 17:12
    I don't think you'll get the closeness of threshold that you want. There's chip to chip variation and variation across the chip due to thermal and voltage gradients, let alone induced voltages as different portions of the chip become active. You're asking for control of the threshold on the order of uV which you won't get.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2012-07-11 18:25
    Unless you need simultaneous conversions, you could use an analog multiplexer feeding a single sigma-delta ADC. (That's the way it's done in the S2 robot, BTW.)

    -Phil
  • jmgjmg Posts: 15,183
    edited 2012-07-11 18:30
    lonesock wrote: »
    ...if any 2 pins on the prop had the exact same input voltage threshold (i.e. within a few uV)...does anyone know if this is the case?

    Some milli volts, perhaps, but Micro volts ? - think about it for a moment. Why do you want uV ?

    Even op-amps which are designed for careful matching struggle to go under 1mV, and if you look at the Comparators that come in small Micros, they specify something like Offset voltage < 20mV - so that is the typical FET matching you can expect.
  • lonesocklonesock Posts: 917
    edited 2012-07-11 20:05
    Thanks, everybody.

    So, here's the reason. I'm measuring a thermocouple, and it's very doable using a sigma-delta ADC, however, the bias resistor to center the input around 0V needs to be very precise (or adjustable, which I dislike). Or the alternative I used, I use another pin with a RC low-pass to shift the negative node of the thermocouple, but I can't just use 50% DUTY cycle, since the comparator threshold isn't right at 3.3/2. In that case I need to push around the value to zero out the reading at room temperature. However, I can also use a 4th pin as feedback as Tubular suggested, IIF the thresholds are very very close to each other. I believe type K thermocouples' sensitivity is on the order of 40uV/C, hence my over-the-top requirements [8^)

    One other possibility, if the thresholds are very similar, is to use a ladder RCtime to read the thermocouple. The 2 nodes will be separated by the delta-V of the thermocouple, and I can time the A and !B while discharging a cap. (does that make sense?)

    thanks,
    Jonathan
  • jmgjmg Posts: 15,183
    edited 2012-07-11 20:35
    lonesock wrote: »
    I believe type K thermocouples' sensitivity is on the order of 40uV/C, hence my over-the-top requirements

    You really need to amplify that first, then run an ADC.

    PhiPi did some great DAC tests here

    http://forums.parallax.com/showthread.php?139705-Is-there-a-recommended-external-DAC-that-works-well-with-the-prop/page3

    These show a DAC (which is one half of the ADC - you have to add threshold noise too...), is comfortably good for 12 bits, maybe pushed to 13 or even 14 bits tops, and LSB errors in the region of 0.4mV can be expected, just in the DAC.

    That's actually quite good for a Digital CMOS part, but not so great relative to your wish of 40uV / 'C.

    So you really need a chopper stabilized amplifier, Microchip have some good examples
    http://www.microchip.com/ParamChartSearch/chart.aspx?branchID=110152&mid=11&lang=en&pageId=79
  • lonesocklonesock Posts: 917
    edited 2012-07-11 20:52
    Yep, final design will use an appropriate part...what I'm interested in now is quick & dirty results...temp range will be on the order of 300 C, so it's not like I need +/- 1 C accuracy.

    Jonathan
  • Tracy AllenTracy Allen Posts: 6,664
    edited 2012-07-11 21:14
    Recall this thread, post #87 in Troubles with Sigma-Delta ADC. Rayman got a thermocouple working (to a degree :thumb:).
  • jmgjmg Posts: 15,183
    edited 2012-07-11 23:38
    lonesock wrote: »
    Yep, final design will use an appropriate part...what I'm interested in now is quick & dirty results...temp range will be on the order of 300 C, so it's not like I need +/- 1 C accuracy.

    A low drift op amp and two resistors is surely Quick, and not nearly as dirty... ?

    Why not target sub 1'C, it should be easy to get, when you have the right parts.
  • Mark_TMark_T Posts: 1,981
    edited 2012-07-12 04:22
    Mike Green wrote: »
    I don't think you'll get the closeness of threshold that you want. There's chip to chip variation and variation across the chip due to thermal and voltage gradients, let alone induced voltages as different portions of the chip become active. You're asking for control of the threshold on the order of uV which you won't get.

    Indeed to get that degree of matching in an opamp takes laser trimming and very clever layout to compensate for thermal gradients - you pay money for the privilege! I'd expect 100mV or so.
  • Mark_TMark_T Posts: 1,981
    edited 2012-07-12 04:38
    jmg wrote: »
    You really need to amplify that first, then run an ADC.

    Or wire 100 thermocouples in series to multiply the voltage!
  • jmgjmg Posts: 15,183
    edited 2012-07-12 05:14
    Mark_T wrote: »
    Or wire 100 thermocouples in series to multiply the voltage!

    Sometimes the best answers are so obvious, we wonder why no one every thought of them before !! ;)
  • Tracy AllenTracy Allen Posts: 6,664
    edited 2012-07-12 12:36
    Jonathan, were you looking at something like the following?
    t-couple-prop.png


    Duty mode provides reference. Sigma-delta ADC. Transmission gate closes, shorts out thermocouple, for auto-zero adjustment. Full scale range about 12mV, and minimal current through the thermocouple. Reference cold junction temperature via a digital sensor.
    t-couple-prop.png
    735 x 423 - 25K
  • Mark_TMark_T Posts: 1,981
    edited 2012-07-12 16:10
    A thermocouple can push a surprisingly large current if you short it, that transmission gate had better have a very low impedance!
  • Tracy AllenTracy Allen Posts: 6,664
    edited 2012-07-12 21:35
    That's a good point about the current. (I usually deal with tiny thermocouples.)

    Maybe it would take an arrangement of switches to connect either the thermocouple, or a short circuit, or an open circuit. A kind of chopper arrangement so that the signal can be moved away from DC. The sigma-delta is a lot better at dealing with AC or stepped signals.
  • jmgjmg Posts: 15,183
    edited 2012-07-12 21:45
    Tracy Allen wrote: »
    That's a good point about the current. (I usually deal with tiny thermocouples.)

    Maybe it would take an arrangement of switches to connect either the thermocouple, or a short circuit, or an open circuit. A kind of chopper arrangement so that the signal can be moved away from DC. The sigma-delta is a lot better at dealing with AC or stepped signals.

    Once you do that, a simple reversal allows alternating readings, with a difference result.

    (This is how chopper stabalised ampifiers work.)
  • Tracy AllenTracy Allen Posts: 6,664
    edited 2012-07-13 11:26
    The switches would add more complexity than it's worth. It'd be much easier to add a dedicated sigma-delta ADC. I've been using the ADS1115 (16 bit with PGA). TI now offers the ADS1118, which is meant specifically for thermocouples and includes the internal temperature reference.
Sign In or Register to comment.