*shrug* I found it useful as source of some inspiration.
That's true. I tried to figure out how multimeters handle overvoltage/range switching on the ADC; some use zener diodes, but I'm not too confident with those. Also, the Fluke 27 schematic didn't shed any light on that either…
That's true. I tried to figure out how multimeters handle overvoltage/range switching on the ADC; some use zener diodes, but I'm not too confident with those. Also, the Fluke 27 schematic didn't shed any light on that either…
The idea is to use clamping diodes so it doesn't go waaay out of range of what the opamp can handle. Then during integration you notice the input is out of range so you switch to a higher range, or open the switch or whatever is the appropriate action. When autoranging voltage I'd say when you detect out of range, you step to highest range, and then step down again to what is appropriate.
And the idea behind the above is that the clamping diodes only have to work for a limited time so you don't stress things too hard. The main drawback (AFAIK) behind those diodes is the leakage current during normal operations.
Generally with a current limiting input resistor and clamping to the power rails (this can be done with diodes, but BJTs and JFETs are also often "abused" for this because they can have lower leakage, also a JFET mux does it for free).
When autoranging voltage I'd say when you detect out of range, you step to highest range, and then step down again to what is appropriate.
On good meters upranging is faster than downranging, because it can detect without a full ADC cycle. And a good meter must withstand it's rated voltage on the lowest range continuously. Autoranging can be switched off. So I see no point in stepping down from the highest range.
When autoranging voltage I'd say when you detect out of range, you step to highest range, and then step down again to what is appropriate.
On good meters upranging is faster than downranging, because it can detect without a full ADC cycle. And a good meter must withstand it's rated voltage on the lowest range continuously. Autoranging can be switched off. So I see no point in stepping down from the highest range.
I see the point in stepping down if it's a DIY approach. I was making the statement within the DIY context, not the professional goodies context. Especially when you are not 100% sure yet about getting everything right, then stepping to highest range first and then slowly (while also not being in a hurry) step it down. Because not blowing shit up is more important to me than a few seconds of productivity gain. After you gain more confidence with your DIY solution you can always change the ranging algo, since that part of the software is not all that complicated.
edit: curse you typos, cuuuurse yooouuuuu
Ok, I see your point. But what is about manual ranging then? Which is the bigger problem.
btw: I would limit a DIY DMM to 200-300V. No need to build a 1000V one.
I have checked the other OSS multimeter schematic, they're using low-leakage BAS116 diodes to effectively limit the overrange to 2.7V, which is appropriate for a 2.5V ADC input. Need to do some spice simulations to see if they handle transients well.
I think it's not a good idea to put the ADC directly on the input. ADCs usually want a low impedance source. And a multimeter should have a high impedance input (10Mohm or more). You need an input amplifier. Then the ADC does see the transients.
That's correct, however, the input buffer can experience the transients as well (we call that "Impedanzwandler" in german, no idea if there's a direct translation).
Generally with a current limiting input resistor and clamping to the power rails (this can be done with diodes, but BJTs and JFETs are also often "abused" for this because they can have lower leakage, also a JFET mux does it for free).
Any
specific examples of both parts and exact use of said parts to achieve this?
Generally with a current limiting input resistor and clamping to the power rails (this can be done with diodes, but BJTs and JFETs are also often "abused" for this because they can have lower leakage, also a JFET mux does it for free).
Any specific examples of both parts and exact use of said parts to achieve this?
Look at the service manuals of older bench DMMs, e.g. HP 3456A, Fluke 8840A, Keithley 19x. I attached one example from the 3456A. Just four 27k resistors and two JFET diodes for clamping. These JFET are ultra low leakage, like 2N4117A.
Edit: The relay is just for range switching.
Generally with a current limiting input resistor and clamping to the power rails (this can be done with diodes, but BJTs and JFETs are also often "abused" for this because they can have lower leakage, also a JFET mux does it for free).
Any specific examples of both parts and exact use of said parts to achieve this?
Not sure if this is what you're after but Dave did a superb video on DMM input protection.
No, that doesn't contain what I was asking about... But in the meantime I did find an answer to the question, which is to use a JFET like the 2N4117A.
Bit expensive though.
Generally with a current limiting input resistor and clamping to the power rails (this can be done with diodes, but BJTs and JFETs are also often "abused" for this because they can have lower leakage, also a JFET mux does it for free).
Any specific examples of both parts and exact use of said parts to achieve this?
Look at the service manuals of older bench DMMs, e.g. HP 3456A, Fluke 8840A, Keithley 19x. I attached one example from the 3456A. Just four 27k resistors and two JFET diodes for clamping. These JFET are ultra low leakage, like 2N4117A.
Edit: The relay is just for range switching.
Lol, why didn't I notice this before? Oh wait, you edited. Or I am just blind. No matter... I got the 2N4117A from this
Texas Instruments Burr-Brown appnote:
http://www.ti.com/lit/an/sboa058/sboa058.pdfBut now I've got double confirmation, which is always good.
At any rate, given the cost I think I will stick with the BAS116 for now and when the rest is proven use the 2N4117A in the next iteration.
Edit: checked an open tab ... yup, I am blind. You already posted that 2N4117A when I replied to jucole. Need glasses.
Maybe you know any good price/performance MOVs as well? Or do I first have to search for those myself, and then 10 seconds before I find them you post the very part numbers I find, 100% as per Murphy's Law or some derivative?
Tradition dictates that you don't know any good price/performance MOVs so I will have to look for them myself.
This would be the first project I design MOVs into, so noooo idea as of yet how to find the "good ones".
No, that doesn't contain what I was asking about... But in the meantime I did find an answer to the question, which is to use a JFET like the 2N4117A. Bit expensive though.
I found these threads insightful :
https://groups.google.com/forum/#!topic/sci.electronics.design/HNi5Bb-rRbEhttps://groups.google.com/forum/#!topic/sci.electronics.design/D53KlEoxmD8If you want to use discretes I'd use one of the low leakage Central Semiconductor diodes (Mouser has them). You won't do significantly better by abusing discrete JFETs or BJTs. As the other thread points out though, some of the integrated MOSFET opamps can do a lot better still.
All that said, if you want to do auto-zero calibration then you need a mux, even if not muxes are cheaper and more compact than lots of relays ... the mux will probably have protection diodes built in.
No, that doesn't contain what I was asking about... But in the meantime I did find an answer to the question, which is to use a JFET like the 2N4117A. Bit expensive though.
I found these threads insightful :
https://groups.google.com/forum/#!topic/sci.electronics.design/HNi5Bb-rRbE
https://groups.google.com/forum/#!topic/sci.electronics.design/D53KlEoxmD8
Thanks! Those are indeed quite informative.
If you want to use discretes I'd use one of the low leakage Central Semiconductor diodes (Mouser has them). You won't do significantly better by abusing discrete JFETs or BJTs. As the other thread points out though, some of the integrated MOSFET opamps can do a lot better still.
Do you have any specific part numbers of the ones you have in mind here, so I can stare at the datasheet?
Edit: search at mouser is not very useful for this one. "Parametric Search for Surface Mount-Diode-Ultra Low Leakage" at the centralsemi website does turn up some. Maybe you mean the CMAD6001 and such?
All that said, if you want to do auto-zero calibration then you need a mux, even if not muxes are cheaper and more compact than lots of relays ... the mux will probably have protection diodes built in.
Same question really, part number to enable datasheet staring?
The thread mentioned the CMPD6001S, but just look up all the central semiconductor diodes on mouser there are a couple with similar specs.
As for muxes, in another thread someone mentioned the MAX4051 since Maxim gives a typical leakage current of 2 fA ... that said, Ti/On semiconductor have versions of the 4051 at a fraction of the cost with a typical 10 fA in the datasheet.
The thread mentioned the CMPD6001S, but just look up all the central semiconductor diodes on mouser there are a couple with similar specs.
Thanks for the confirmation. Search on centralsemi.com turns up basically CMxD6001...
As for muxes, in another thread someone mentioned the MAX4051 since Maxim gives a typical leakage current of 2 fA ... that said, Ti/On semiconductor have versions of the 4051 at a fraction of the cost with a typical 10 fA in the datasheet.
So not the
CD4051B then?
So not the CD4051B then?
10 fA typical at room temperature.
So not the CD4051B then?
What are your requirements? Voltages, resolution, ...?
A CD4051 should be fine for a 3-4 digits meter and probably a disaster for a 6.5 digits one.
10pA more like it, but clean the package, clean the board and have soldermask and guard traces and it is likely to go a lot lower. Drop supply voltage to 15V and it will be worse, as this is likely specced at the best conditions.
So not the CD4051B then?
What are your requirements? Voltages, resolution, ...?
A CD4051 should be fine for a 3-4 digits meter and probably a disaster for a 6.5 digits one.
I wasn't planning on using the CD4051B. I just linked to it to check where Marco got those awesome leakage figures for a 4051 device. But an
oopsie of 3 orders of magnitude would explain things.
10pA more like it, but clean the package, clean the board and have soldermask and guard traces and it is likely to go a lot lower. Drop supply voltage to 15V and it will be worse, as this is likely specced at the best conditions.
Also pre-assembly cleaning? Or just the whole assembled board? If pre-assembly, any handy tips? Usual procedure would be to put it in an IPA filled ziplock bag in warm water in an ultrasonic bath, but maybe there's better ways?