Keyboard Shortcuts
ctrl + shift + ? :
Show all keyboard shortcuts
ctrl + g :
Navigate to a group
ctrl + shift + f :
Find
ctrl + / :
Quick actions
esc to dismiss
Likes
Search
DM501A Input Leakage Source
|
I have a pair of DM501As that I use as my every day go-to DMMs on my bench. They work well. I have noticed, though, that with an open input on the 200mV range, one reads about +11mV, and the other about +3mV. Goes to zero when the input is shorted. On the 11mV unit, this corresponds to about 1.1nA on board leakage into the A/D input node and the associated 10Mohm input resistance. When I isolate the A/D input and ground it through 10Meg, the reading is around 0.01mV. So it is not the A/D chip itself.
The schematic shows a chain of switches in the S1 assembly, between the input attenuator assembly and the A/D. By watching the display as I switch ranges and short the input, it appears that the leakage injection is into that chain of switches. I isolated the relatively long pc track from the last switch (S1-N) to the A/D (R1613) and it looks like the injection is somewhere in the S1 assembly itself. I sprayed PC cleaner in, around and through the switches liberally, but after drying it - same result. But mere dirty switches aren't enough - there has to be a source of +V somewhere close by. To isolate possible switch leakage from other parts of the circuit, I have grounded, in turn, all of the feeds from the ohms, AC and temperature circuits at the respective switch terminals, with no difference to the display. I can't readily see any other sources. And now I am out of ideas. So I am appealing to the wealth and breadth of knowledge in this forum for any suggestions. What else can I do to identify the leakage source? Is switch removal REALLY as simple as the manual suggests "...by carefully unsoldering the connections and pulling....". Dozens of pins have to be completely cleanly unsoldered for this to work... I really do not want to go there just yet. Yes, I realise that this parameter is not specified in the manual, and this effect does not detract from the usability of the instrument. It's because I can :-)! Roman |
|
Don't worry about it too much. Any input amplifier of any system needs to have a certain DC input bias current for operation. It may be very small, but won't be zero, or constant, over operating conditions. Whenever the input is scaled beyond its native range with resistive attenuators, it should read very close to zero with no input connected. Inside the native range, the bias current still needs to flow, but can't go into any attenuator resistance (open circuit), so the voltage indicated is the bias current times the leakage resistance to anything and everything around, including non-zero potentials. Some DMMs always have some attenuation or termination resistance (like 1 meg or 10 meg), so you usually won't see much effect. If your DMMs have a definite, fixed input resistance in all ranges, yet still read a definite voltage, then there is some leakage from somewhere, or a definite bias current large enough to show.
If the amount is "in spec," then these various leakages were anticipated and specified accordingly. If you suspect the switching system, then look at all the switch poles and circuit board runs for any adjacent the signal path, or even far removed. This may not be apparent from a single schematic view - range and function switches often have logical coding for decimal points, relay drive, and such, that may show up on different schematics. It doesn't have to be a supply voltage, but anything with a potential away from ground, under any condition. For instance, the input of a TTL gate is a voltage source, not just a passive item. Ed |
|
Hi Roman
If despite Eds argument you need to fix the switches. There is a document here that explains how to open the switches and clean them. I do have a few DM501A that are waiting to be cleaned and I would like info on how to lubricate them properly after assembly and what lubricants to use. Regards Gudjon |
|
First, what Ed said. :)
toggle quoted message
Show quoted text
Second, you're looking for leakage on the board, but you won't find it there, so don't tear things apart in a hunt for what is a ghost. The leakage is coming from the protective clamping diodes and/or the gate of the input FET. For the latter, one can expect a gate leakage current that is very, very approximately, roughly 6-7 orders of magnitude below IDSS. So, a 10mA IDSS (a reasonable guess for the input FETs) gets you leakage of the order that you are inferring. Third, what Ed said. --Cheers, Tom -- Prof. Thomas H. Lee Allen Ctr., Rm. 205 350 Jane Stanford Way Stanford University Stanford, CA 94305-4070 On 4/19/2022 21:56, romeo987 wrote:
I have a pair of DM501As that I use as my every day go-to DMMs on my bench. They work well. I have noticed, though, that with an open input on the 200mV range, one reads about +11mV, and the other about +3mV. Goes to zero when the input is shorted. On the 11mV unit, this corresponds to about 1.1nA on board leakage into the A/D input node and the associated 10Mohm input resistance. When I isolate the A/D input and ground it through 10Meg, the reading is around 0.01mV. So it is not the A/D chip itself. |
|
Roman,
I agree with what Tom Lee mentioned. I can¡¯t remember any meter in the lab (over a dozen) including brand new units from various manufacturers that doesn¡¯t show some minor offset with the input folating. Remember what type of input impedance you are dealing with that is floating without anything connected. Have you performed calibration of these units? If you will notice in the manual Tek instructs you to connect a short between the input high and low jacks and then adjust the internal zero control for a reading of 0000 on the display. So from this it is somewhat alluded to that Tek knows that there will be a possible inherent offset reading if you don¡¯t short the input. If you decide to pursue this issue I am sure that it will become entirely academic in nature. You may find some stray leakage paths but most of it will be attributable to artifacts in the analog front end. The 501As in the lab show the same offsets that vary somewhat between Besides, if you are concerned with levels to femtoamps and picovolts with this item you might be using the wrong meter. There are far better unit out here that exceed a 4-1/2 digit multimeter. I find the Fluke 8588A 8-1/2 digit meter suitable for applications such as this. Greg |
|
Tom,
Thanks for your input. I do note that you are very generous with your advice on this board - as are many others. I really appreciate it. In this case, I think your first and third points are well made. As to the second, that is almost certainly true of a normal scope amplifier input, but in this case there are no such devices ¡°left¡± of the A/D chip itself, which I have absolved by test as the cause. Roman |
|
Greg,
Thanks for your response. I realize that this leakage does not affect the utility of the meter one whit. I am just trying to improve what I have, because I can. Academic, of course, but interesting. I am just a retired old bloke who loves to play. Not particularly interested in measuring fA and pV, but only in making my two instances of this meter as good as they can be. And I am intrigued that you appear to be in a lab that has 70¡¯s vintage 501As alongside Fluke 8588s. Says a bit for the longevity of this Tek stuff. I guess that¡¯s why we love to play with it! Just out of curiosity, what sort of offset do you see on yours on 200mV DC range, OC input? Is it always positive? Roman |
|
Hi Roman,
toggle quoted message
Show quoted text
Just for grins, what happens if you touch the input terminal of the plug-in while in dc volts mode? Don't touch ground the ground terminal at all -- just be an antenna. If the reading rises, try connecting a capacitor (say, some nanofarads) across the input terminals of that plug-in. A small amount of line noise pickup can be aliased down to dc by dual-slope converters. This test will tell you whether that's happening here. Tom -- Prof. Thomas H. Lee Allen Ctr., Rm. 205 350 Jane Stanford Way Stanford University Stanford, CA 94305-4070 On 4/21/2022 06:24, romeo987 wrote:
Tom, |
|
Roman,
I¡¯m seeing somewhat less compared to what you see on the 200 mV DC range. And it varies between plug-ins with one being ~6 mV highest and the lowest being ~2 mV. Of course shorting the input zeros out the reading. What is interesting is that when placing the EXT/INT switch on INT the offset minimizes. A quick look at the schematic shows several functions being switched in different parts of the circuit aside from merely transferring from the front panel jack to the rear card edge connector finger. I don¡¯t have time right now to investigate why. I also have a couple of DM502As and threw them in a TM frame. They read 0.1 and 0.2 mV with the input open (not shorted). Pushing the EXT/INT switch didn¡¯t change the reading either nor did shorting the input. So I suspect that they need to go through the zero calibration process spelled out in the manual. The manual provides some helpful block diagrams showing switching and signal routing that may assist you with your investigation. And one other thought - have you checked for any ripple in the DC supply lines from the TM frame power supply just to make sure that it isn¡¯t helping this issue? And how clean are the card edge finger contacts? A slightly elevated ground condition through dirty fingers can cause many strange artifacts as well. Greg |
|
Greg, Tom, Ed
Thanks for your inputs, guys. Tom, I tried your antenna thing but it made no visible difference, Ed, I worked on your idea of inter-switch leakage. You are right, there is a lot of +5V switching going in in poles physically adjacent to the sensitive input node. In particular, switching back and forth rapidly between Temp and 200mV had an appreciable (temporary) effect, so I liberally squirted pcb cleaner into that particular switch. After drying out, the offset was reduced from around 11mV to around 4mV. This suggests that the cause, at least, has been identified, if not the complete solution. Given what Greg reported in his units, and the closeness of so many 5V switching nodes, this is probably as good as it gets. Looks like the designers didn¡¯t try too hard to guard the input node, but that¡¯s probably a bit harsh coz of course 400pA is probably not too shabby for this class of DVM. That Tek article that Gudjon refers to in his post is great if the switch is on its own on the bench, but quite impossible if the switches are still in the instrument. Maybe I¡¯ll have another go one day¡.. Thanks again, guys. Roman |
|
Hi all
Roman: I am pretty sure that by removing the trim capacitors, which is easy, you can disassemble and clean all the switches and assemble them again. I will try this as soon as I get some time but I cannot promise that will be any near time but I promise to take pictures. My working DM501A shows 0.2mV when unconnected on the 200mV range. It has PCB's with green silkscreen on. There are versions without silkscreen on and I am wondering if they leak more. Does your DM501 have silkscreen? ? But since this is being discussed. Does the contact oil increase leakage between the switch pins? Does anyone have advice on what contact oil/grease to use on the switches? Regards Gudjon |
|
Gudjon,
I¡¯ll be interested to see how you go - as far as I can see, the switches are tucked well in under the attenuator pcb and accessing the rear of them would require quite some dexterity! I¡¯ll be very interested to see how you go. In any case, my culprit appears to be the function switch, rather than the range switch, which looks even less accessible. My pcb does not have a green silk screen, but during my testing I isolated the pc traces (although possibly not all of them) without effect. There is some sort of clear covering over the pc traces, though - I need to push the test probe quite hard to make contact. Roman |
|
Um, perhaps you mean green soldermask.? Soldermask is most often green in color, although red, blue, and purple are becoming more popular lately.? Most PCB fab shops would rather do everything the same, though, and you may pay extra for anything other than green.Silkscreen lettering is most often white.? Might see yellow from time to time.? Both are chosen for contrast against the relatively dark green soldermask over copper or epoxy-glass board material.? Some RF/microwave/high speed digital board material is close to white, so black or other dark soldermask may be appropriate there.?Jim FordSent from my T-Mobile 4G LTE Device
toggle quoted message
Show quoted text
-------- Original message --------From: romeo987 <romeo987@...> Date: 5/1/22 5:06 PM (GMT-08:00) To: [email protected] Subject: Re: [TekScopes] DM501A Input Leakage Source Gudjon,I¡¯ll be interested to see how you go - as far as I can see, the switches are tucked well in under the attenuator pcb and accessing the rear of them would require quite some dexterity! I¡¯ll be very interested to see how you go. In any case, my culprit appears to be the function switch, rather than the range switch, which looks even less accessible.My pcb does not have a green silk screen, but during my testing I isolated the pc traces (although possibly not all of them) without effect. There is some sort of clear covering over the pc traces, though - I need to push the test probe quite hard to make contact.Roman
|
|
I figured, Roman.? But it's good to get the information out there for those who haven't been designing boards for decades.JimSent from my T-Mobile 4G LTE Device
toggle quoted message
Show quoted text
-------- Original message --------From: romeo987 <romeo987@...> Date: 5/2/22 3:47 AM (GMT-08:00) To: [email protected] Subject: Re: [TekScopes] DM501A Input Leakage Source Jim,Yes, brain f**t, of course solder mask.Roman
|
|
Hi Roman
Now I have tested cleaning a 20V range switch without unsoldering it. I had to mill a U-shaped plastic piece to fit the connector sleeves into to be able to reassemble the switch. The 200mV, 2V and 20V range switches can be disassemled without removing any other components but to fix the other range switches, the AC compensation capacitors will have to be removed. After having cleaned the switch several times it was not stable and the value changed if I pressed it hard. This turned out to be a broken soldering behind the Caddoc resistor. After soldering it, the module works like charm. Just needs to be calibrated. I have three DM501a multimeters. The offset voltages when open are 0.2, 0.5 and 1 mV. All of the modules have PCB's with soldermasks (sorry for my silkscreen mistake). According to this article, the soldermask helps to keep moisture out of the PCB. But the leakage may also stem from dirt in the switches. I will clean the 200mV switch on the meter with 1mV offset and see if it helps but I won't have time for a few weeks. /Gudjon |
to navigate to use esc to dismiss