EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: jars121 on August 25, 2024, 07:13:55 am
-
Afternoon all,
I'm designing a circuit to interface with K-, T-, E- and J-type thermocouples. Most of the heavy lifting is done in the Sigma-Delta ADC, which has an internal PGA set to 8x (the maximum gain available) and a Vref of 2.5V, providing a FSR from -0.15625V to +0.15625V. Even with this relatively small FSR, the narrow range of the T-type thermocouple in particular means that I'm not using much of the FSR if digitising the thermocouple signal directly (range of the T-type is -6.258mV to 20.872mV, which represents only ~8.6% of the available ADC FSR.
Rather than provide a different gain for each thermocouple type, I'd like to use a fixed gain approach, that uses as much of the ADC FSR as possible for the K-, J- and E-type thermocouples (amplifying their output such that no clipping occurs on the top end of their range), and accept that the T-type thermocouple range will be 'good enough'.
With that said, I think a gain on the front-end amplifier of ~2.2 will be suitable for this purpose.
I've looked at some instrumentation amplifiers, such as the AD8230, but note that small gain values (<10) are outside their 'typical' operation conditions.
Can anyone recommend a low-offset, low-drift, 'low noise' amplifier (operational and instrumentation, whichever is more suitable)? I have 6 discrete thermocouple input channels, so a $10+ instrumentation amplifier might be a little too expensive.
Thanks!
-
I recently replied to a guy's post on Reddit - he was making a circuit board with 8 x MCP9600 : https://www.microchip.com/en-us/product/mcp9600 (https://www.microchip.com/en-us/product/mcp9600)
The chip's kind of expensive at 9-10$ a piece, so I suggested the possibility of using a couple analogue 8:1 switchers to switch the input of the converter between probes.
If you don't need very fast update rates, this may work for you as well.
I was suggesting switches like TMUX1108 : https://www.digikey.com/en/products/detail/texas-instruments/TMUX1108PWR/9861450 (https://www.digikey.com/en/products/detail/texas-instruments/TMUX1108PWR/9861450)
This has a typical resistance of around 2.5 ohm with a maximum of 4-5 ohm , and small variation between channels of around 150 mOhm
A also pointed to a better switcher, NX3L4051PW-118, but which can handle only up to 4.3v : https://www.digikey.com/en/products/detail/nxp-usa-inc/NX3L4051PW-118/3679433 (https://www.digikey.com/en/products/detail/nxp-usa-inc/NX3L4051PW-118/3679433) (this one has a typical switch resistance of only around 0.5 ohm)
so if your microcontroller and mcp9600 runs on 5v, you'd need a small 3.3v voltage regulator and/or some level shifting for the inputs of the switchers - an i2c port expander or shift register would work (you need 4 inputs per analogue muxer, the 3 bits to represent 0-7 value and the enable to make the switch, so with two analogue switches you'd need 8 inputs. As the mcp9600 is already on i2c, you could use a i2c shift register/port expander to control the switchers)
The MCP9600 has a 18 bit ADC and a measurement time of 320ms at 18bits (+20ms for the temperature calculation), but can be switched to lower resolutions
Sampling Rate (TA = +25°C)
tCONV — resolution
— 320 — 18-bit resolution
— 80 — 16-bit resolution
— 20 — 14-bit resolution
— 5 — 12-bit resolution
Temperature Calculation Time t CALC — 20 — ms TA = +25°C
so if you want fast updates, I imagine you could loop through the 6 and in each loop measure only one input at the full 18 bit, and the others at lower 16 bit resolution - you'd measure 1 x 340ms + 5 x 100ms = 850 ms or so to measure all 6 channels with this chip.
-
Thanks for that mariush, I appreciate your input.
I had evaluated that available thermocouple interface and temperature-to-digital IC options thoroughly before setting down this path. Unfortunately I need considerably higher sampling rates than offered by any of these integrated solutions (up to 1kHz sampling for some of the finer gauge thermocouples).
As such, I'll be using a discrete ADC, and am looking for a suitable front-end amplifier to better utilise the available FSR.
-
Most of the heavy lifting is done in the Sigma-Delta ADC, which has an internal PGA set to 8x (the maximum gain available) and a Vref of 2.5V, providing a FSR from -0.15625V to +0.15625V.
...
Rather than provide a different gain for each thermocouple type, I'd like to use a fixed gain approach, that uses as much of the ADC FSR as possible for the K-, J- and E-type thermocouples (amplifying their output such that no clipping occurs on the top end of their range), and accept that the T-type thermocouple range will be 'good enough'.
With that said, I think a gain on the front-end amplifier of ~2.2 will be suitable for this purpose.
Take a close look at the drift and noise specifications of your sigma-delta ADC. At maximum gain, it may be as good as any reasonable amplification stage so little or nothing will improve with external amplification. Because of how sigma-delta ADCs work, their inputs perform about as well as chopper stabilized amplifiers.
I've looked at some instrumentation amplifiers, such as the AD8230, but note that small gain values (<10) are outside their 'typical' operation conditions.
Instrumentation amplifiers are sometimes used in thermocouple applications, but are inherently noisier than singled ended amplifiers.
-
Take a close look at the drift and noise specifications of your sigma-delta ADC. At maximum gain, it may be as good as any reasonable amplification stage so little or nothing will improve with external amplification. Because of how sigma-delta ADCs work, their inputs perform about as well as chopper stabilized amplifiers.
To clarify, are you saying that the additional external amplifier (even with a low gain of 2) won't actually add any useable (i.e. non-noise) signal?
Instrumentation amplifiers are sometimes used in thermocouple applications, but are inherently noisier than singled ended amplifiers.
I'd be open to single-ended, I have no preconception as to the external amplification stage.
-
Take a close look at the drift and noise specifications of your sigma-delta ADC. At maximum gain, it may be as good as any reasonable amplification stage so little or nothing will improve with external amplification. Because of how sigma-delta ADCs work, their inputs perform about as well as chopper stabilized amplifiers.
To clarify, are you saying that the additional external amplifier (even with a low gain of 2) won't actually add any useable (i.e. non-noise) signal?
That is exactly what I am saying if the sigma-delta ADC is operated at maximum gain. At lower gain, which will provide for a higher sample rate, the ADCs input noise is higher so an external amplifier has some benefit.
All of this will depend on what ADC you are using and its configuration. For instance an AD7780 with its PGA set to 128x has an input noise of about 50nVrms (unspecified low bandwidth). Improving on that will require an LT1028 class of operational amplifier, in parallel with a chopper stabilized amplifier to control the drift of the LT1028, which is not an inexpensive proposition.
-
With thermocouples one can likely have one side connected to a fixed ground and use normal 1 sided amplifiers. It needs some care with the ground and maybe ground current compensation to keep a ground shift small.
A suitable low noise AZ amlifiers could be OPA388 or OPA189. I would use an analog more like a gain of 10 or even more, as this can can make it a bit easier to get low noise. There are also ready made resistor pairs for a reasonable stable / accurate gain of 10. A gain of 2 is hardly worth the effort.
There are a few ADCs with pretty low noise, especially those made for a direct DMS interface.
A word of caution to the LT1028 and other very low noise amplifiers. They often have quite some current noise and get the full potential only with very low impedance (e.g. < 30 ohm for the the LT1028). The very fine wire thermocouples may already be higher than that. An OPA210 would a candidate for slightly high source impedance.
If noise is really critical, separate amplifers and ADCs per channel could be an option instead of switching between channels. There are relatively low cost ADC with multiple channels, originally made for power metering.
-
Thank you both for your input.
That is exactly what I am saying if the sigma-delta ADC is operated at maximum gain. At lower gain, which will provide for a higher sample rate, the ADCs input noise is higher so an external amplifier has some benefit.
All of this will depend on what ADC you are using and its configuration. For instance an AD7780 with its PGA set to 128x has an input noise of about 50nVrms (unspecified low bandwidth). Improving on that will require an LT1028 class of operational amplifier, in parallel with a chopper stabilized amplifier to control the drift of the LT1028, which is not an inexpensive proposition.
I'm planning on using the AD7779. I'll be using 6 channels for the thermocouples themselves, and an LTC2997 on the 7th channel for CJC.
With thermocouples one can likely have one side connected to a fixed ground and use normal 1 sided amplifiers. It needs some care with the ground and maybe ground current compensation to keep a ground shift small.
A suitable low noise AZ amlifiers could be OPA388 or OPA189. I would use an analog more like a gain of 10 or even more, as this can can make it a bit easier to get low noise. There are also ready made resistor pairs for a reasonable stable / accurate gain of 10. A gain of 2 is hardly worth the effort.
There are a few ADCs with pretty low noise, especially those made for a direct DMS interface.
A word of caution to the LT1028 and other very low noise amplifiers. They often have quite some current noise and get the full potential only with very low impedance (e.g. < 30 ohm for the the LT1028). The very fine wire thermocouples may already be higher than that. An OPA210 would a candidate for slightly high source impedance.
If noise is really critical, separate amplifers and ADCs per channel could be an option instead of switching between channels. There are relatively low cost ADC with multiple channels, originally made for power metering.
I was giving this design some further thought last night, and I've come to the realisation that I've got the configuration the wrong way around. Rather than max out the PGA gain in the ADC and use such a small gain on the front-end amplifier, I should instead choose a larger front-end amplifier gain and then 'tune' the signal using the ADC PGA gain. I put together some numbers to validate this approach:
(https://www.eevblog.com/forum/projects/thermocouple-amplifier-recommendation/?action=dlattach;attach=2350745;image)
The values in columns B and C are taken directly from thermocouple tables for each of the 4 thermocouple types. I've selected a front-end amplifier gain of 16, and then assigned a per-thermocouple-type ADC PGA gain (between the available values for the AD7779 of 1, 2, 4 and 8 ) to provide the maximum FSR coverage. There are other front-end gain values I can play with the find the sweet spot, but this the general approach I'm leaning towards.
In terms of the front-end amplifier, I had come across the OPA2387 and ADA4523-1 in my research last night. Their parameters seemed to be quite impressive, but the OPA388 and OPA189 came up frequently in my searching as well.
-
The ADA4523 is super low noise, but the rather high supply current and thus heat can be a problem or at least need extra care.
The OPAx387 is a little higher noise than the 388 version, but comes with less power consumption. So it could still be worth it.
More analog gain and less at the ADC makes sense. There is likely no need to overdo things, as the resolution of the SD ADC is usually plenty and using not the full range should not be an issue. Adjusting in 1:2 steps from the ADC internal gain should be well good enough.
-
With thermocouples one can likely have one side connected to a fixed ground and use normal 1 sided amplifiers. It needs some care with the ground and maybe ground current compensation to keep a ground shift small.
The use of instrumentation amplifiers as thermocouple amplifiers comes about in industrial applications where the common mode voltage between ends of a long thermcouple wire may be 10s or 100s of volts. Galvanically isolated thermocouple amplifiers are also used. In a lab environment, there is no reason not to use the lower noise singled ended configuration.
I would tend to use an ADC with differential input to remove microvolt level common mode errors in the signal conditioning circuit itself, but an external operational amplifier can do this as well for a single ended ADC. As Kleinstein points out, there are plenty of error sources in thermocouple signal conditioning so maybe this does not matter.
-
You're exactly right David, this solution may well find itself in an industrial environment, so common mode considerations are definitely a factor.
I found this isolation with gain stage approach in this TI Tech Note (https://www.ti.com/jp/lit/ab/sbaa238a/sbaa238a.pdf?ts=1724705314373&ref_url=https%253A%252F%252Fwww.google.com%252F):
(https://www.eevblog.com/forum/projects/thermocouple-amplifier-recommendation/?action=dlattach;attach=2351775;image)
They've used TLV6002 amplifiers, which I'd probably replace with the OPA388 per Kleinstein's comments below, and I'd probably use the AMC3330 isolated amplifier (instead of the AMC1301 as shown) as it provides an internal isolated DC/DC converter so I can use a single, low-side supply.
I think by choosing a reasonable fixed front-end amplifier stage gain, with per-thermocouple-type ADC PGA gain as outlined in my previous post, this should provide a reasonable solution.
-
The AMC3330 can only provide 1 mA of current as auxiliary power. The OPA388 is to high in current consumption to be powered from the isolator.
When an isolator is used, there is no more real need to also have differential amplification. The differential amplification is still an easy way to create the symmetric differential drive to an ADC / the amplifier. Noise wise and from the power consumption it is not ideal.
The big question for the isolation is if the thermocouples are externally isolated. This is having them directly mounted to the part to measure or have isolated thermcouples (e.g. thermocoax with isolation). The electrical isolation of the TC can add extra delay. So the very fast measurements may well be non isolated. A fast measurement is anyway a rather special case, that also needs extra care with mains hum.
-
The AMC3330 can only provide 1 mA of current as auxiliary power. The OPA388 is to high in current consumption to be powered from the isolator.
When an isolator is used, there is no more real need to also have differential amplification. The differential amplification is still an easy way to create the symmetric differential drive to an ADC / the amplifier. Noise wise and from the power consumption it is not ideal.
Very good point, I hadn't yet gone through the AMC3330 datasheet in detail. There are plenty of similar isolated amplifiers without the integrated DC/DC converter, I was just trying to avoid the need for an external isolated supply if possible, but that isn't difficult to add in.
With regards to your second point; many of the 'thermocouple amplifiers' I've looked at use differential amplification internally (e.g. the AD849x series) so I figured I'd follow suit.
The big question for the isolation is if the thermocouples are externally isolated. This is having them directly mounted to the part to measure or have isolated thermcouples (e.g. thermocoax with isolation). The electrical isolation of the TC can add extra delay. So the very fast measurements may well be non isolated. A fast measurement is anyway a rather special case, that also needs extra care with mains hum.
As you've stated, it will be a little of both. I'm trying to accommodate both approaches as much as possible.
-
The differential amplifier or an isolated input are both ways to handle common mode voltage. There is little need to have both.
The cheap DCDC converters have quite some CM ripple current - this may cause some EMI like issues.
For the isolated input version, there is now also the option to have the isolation on the digital side and thus a separate ADC per channel. So essentially seprate 1 channel versions and combining the digital results. The demands on the ADC and refenrece are usually not that extreme, as thermocouples are not super high accuracy anyway. This avoids the not that perfect analog isolators.
-
I've done some further reading and think I have a better grasp on things now, thank you again for your input!
I understand that having both a differential amplifier stage and an isolated amplifier is redundant. As each of the inputs may be connected to a grounded or isolated thermocouple (completely at the discretion of the end user), I think the 'safest' approach is to use an isolated amplifier as the first stage.
The final question I have is this. I would like to provide some additional amplification after the AMC1400, to better utilise the FSR of the ADC. As the signal has now been isolated, and any common mode noise and/or offsets addressed, can the differential output of the AMC1400 go through a single-ended amplifier stage, or is there still merit in differential amplification? I'd be remiss if I didn't leverage that true differential measurement capability of the ADC, but in this instance it might not be necessary?
-
range of the T-type is -6.258mV to 20.872mV, which represents only ~8.6% of the available ADC FSR.
Have you considered the HX711? Note that you'll need a bias divider and capacitors to ground in order to get the common mode voltage to what it expects and reduce noise.
-
An extra amplifier behind the AMC1400 should be differential. Using the full ADC range should be not that important. The system should be more limited by the accuracy of the TE and the noise of the amplifier, not the noise or linearity of the ADC if a reasonable good ADC is used.
If extra amplification is wanted, this would be better on the input side. The AMC is relatively low input inpedance and quite some bias and thus not suitable to directly measure a thin wire thermocouple that can have a resistance in the 100 ohms range. The noise does not look that great too - so more like not the test choice overall.
The analog isolation amplifiers are a bit tricky from noise, CM injection and accuracy. I would really consider more like separate amplifier and ADC before the isolation. The AMC1400 looks internally like SD ADC, isolation and DAC already.