Tonghui TH2821A LCR Meter

You really did fantastic! That's what opened this whole subject.

I think my original problem started when I tried to operate the unit with a completely dead battery. That may have dropped the voltage to the uP enough to give the erratic results I was getting.

Now I'm thinking as long as there is some charge remaining in the battery, it should work fine while it is charging.

So I think it probably can sit on the shelf for long periods when it is not needed, and still be ready for use when it is required for a new project.

Apparently NiMh shares some of the same features as regular NiCd, and it doesn't particularly care what state of charge it is in. The fast charge cycle that your measurements show probably charges the cell at 100mA, with the other 40mA going to the electronics.

That amount of charge is well within the recommended limits for the battery, so I think the designers have done a good job to give a long life to the battery, and it should not be a concern. New ones are available for $5.00 if you shop around, and some sales might even drop that to $1.50. So I'm not worried about the battery life or replacement cost any more. I think we should just use the unit when it is needed, and leave it on the shelf the rest of the time. [...]

Thanks, John. That would be a special privilege. But recall that Jon Kirwan has a unit, and P E Schoen has the MCP BR2822 which appears to be identical:

So there are some who might be interested in the discussions, and maybe others who might be interested in getting one also. But if we disappear, no one else will be able to find out more about it.

What I would like to do is find out more about how this thing works and why it is so accurate. Is it because of good snr, good circuitry, quality a/d's, software, components, or some other factor. What can we learn from this instrument that can help in other projects?

But there are times when the readings fluctuate for no apparent reason. I'd also like to find out why it does that and what can be done to overcome the problem. The new model has averaging, so that would help. But it probably slows down the measurement, and it also costs much more. Is there any way we can get the same effect without the added cost?

And I certainly would not mind modifying my instrument if if would improve the performance. For me, that is a given.

Someone may come across a piece of information or some data that helps unravel these mysteries, and post it here. But if we go private, this is unlikely to happen.

We are not taking up much room, and we are not bothering anyone, so my vote is to stay here and not worry.

Thanks,

Mike

Sounds good to me.

It fluctuates around zero quite a bit when I check the resistance of an

You make some good, valid points. I agree.

John S

I tried the link and the resulting pdf has a problem and Adobe won't render it. I there another English manual for it?

?-)

I'll buy some of that. These have just about the properties i am looking for. So i am real interested.

?-)

looking

If you want to find bad electrolytics, I don't think you can find a better bargain. It gives the dissipation of the capacitor, so you don't have to fool with lookup tables like you do with a conventional ESR meter to see if the cap is good or bad. Besides giving an easy way to check electrolytics, it also gives the capacitance value to four digits. No ESR meter does that. It also measures inductors and resistors, which no ESR meter can do. It has settable upper and lower limits for bin sorting. And it's less expensive than most ESR meters that I could find.

I just checked eBay. On Jan 14, 2012, the lowest price was $159.99 with $25.00 shipping. Another is available at C $187.99 and C $12.00 shipping, so you have to check both carefully for the best deal.

I'd also get the TH26029 SMD Test Tweezer Probe for $29.99.

The newer model has averaging, but it is closer to $500. I don't think it's worth considering. Once you start getting into those numbers, might as well consider going to a more expensive bench model with greater accuracy and frequency range.

If you order one, make certain to get a 120VAC adapter instead of the

Hope this helps.

Mike

Update : it actually displays 4 digits plus a decimal on large electrolytics. But the two least significant digits dance around a lot.

I switched to a 10uF 250VAC +/-5% film capacitor, and it read 10.045uF with the last digit fluctuating between 4 and 6. So I conclude electrolytics are not very stable. But the readings are quite consistent on a quality part.

I've been monitoring the battery self-discharge. It is terrible. It loses a significant fraction of the charge in a week. Leaving it plugged in has no effect. It does not trickle charge while it is turned off.

One solution might be to simply leave it plugged in a turned on. It says "Auto Power Off" under the display, so that might not work. But it seems to stay turned on when it is measuring something, like a large electrolytic.

I found it really is necessary to turn it off when moving from one AC plug to another, otherwise it may give strange results. In fact, it might be a good idea to post the operating notes found on page 5 of the manual:

Notes on Use

1. This meter is only for indoor use.
2. Turn off the TH2821A while switching the power supply between battery and DC adapter or replacing the battery.
3. Although internal circuit protection is provided, DC voltage or current may damage TH2821A. Before you measure a capacitor, be sure the capacitor is fully discharged.
4. Charging may disturb measurement result sometimes.
5. Nickel Metal Hydride rechargeable battery can be used for the power supply. TH2821A will not work normally when battery voltage is less than 6V.
6. The 12V AC to DC adaptor is recommended to be used for TH2821A power supply.
7. Perform Open and Short corrections for accurate measurement especially when test fixture is changed.
8. The functions locked with password are not accessible by users.

My biggest problems have been from ignoring rules #2 and #4. I used to freak out when it didn't do what I thought it should be doing, but after following the notes, it works fine. And I will add protection against charged caps, as stated in rule #3. I found that human error can quickly wipe out a good instrument when you are working with caps charged to high voltage.

Now I'm wondering if a plain 9V NiCd can be used in place of the NiMH battery. They are supposed to be interchangeable, but the NiCd may have a capacity of 120mAh, where the NiMh has 200mAh. So it may last 3 hours on NiCd, and 5 hours on NiMh. The difference is not significant for me.

But the NiCd has a much lower discharge rate, so you wouldn't be running into the issue of leaving it on the shelf for a month, then trying to use it when the battery is completely discharged. I'm tempted to get a 9V NiCd and see how it works. If it solves the self-discharge problem, that's the way to go.

So now I'm happy with the stability of the meter readings on a quality capacitor, and I won't worry about fluctuation on electrolytics.

But I'm very unhappy with the self-discharge of NiMh. That may be solved by switching to a plain NiCd, so it is more likely to have plenty of charge remaining when I want to use it for some quick measurements. Afterwards, I can just leave it plugged in and turned on to bring the battery back up.

I think that covers the issues up to now on this instrument. It is still by far the best bargain for LCR measurements I have found anywhere.

Mike

Try different frequency settings if you have not already

Are you kidding? Fluctuations of several millifarads out of 10 uF is a dream come true for me.

I just charged mine from a lab supply and found it to respond as before, that is, about 3.7 mA in an OFF condition and about 145 mA in an ON condition. It was at about 1 bar of the battery indicator on the display when I started. About an hour later the current dropped to 45 mA as before and the display indicated 3 bars.

I do not seem to have a problem with self-discharge. I'm wondering what's going on. Maybe I should open mine to see if the battery is like yours. Maybe you should contact them with your situation and ask for assistance.

Also, I've learned not to trust inductance measurements of a 60 Hz transformer even at 100 Hz. Maybe because it is seeing lots of leakage inductance or iron loss or something.

I agree that it is a great bargain. We just need to learn about its limitations. However, your battery problem bugs me. Do you suppose it is not turning off with the power button?

John

Yes, that seems to have little effect on the electrolytics.

I think you meant fluctuations of several nF. Yes, it has drifted up to 10.048uF, and is bouncing around between 10.047uF and 10.049uF.

But it spends most of its time on 10.048uF.

I tried to take a picture to show you,but ironically the LiIon battery in my camera is dead so I have to wait to charge it back up.

I'll post pictures of the display and the capacitor tomorrow. I expect the reading will change a little, but that's jut normal drift. Not to worry.

I charged mine fully then plugged in but left it turned off for about a week. When I moved it to a different outlet and turned it on, the battery indicator showed only one bar.

Either the battery status indicator is not accurate, or the NiMh has a serious self-discharge. I think it's the NiMh.

I didn't find any problems on the inductors. I'll check more later.

Could be. The operating notes specifically mention turning the unit off while moving to a different outlet. I did that. The battery did not last long.

But all the references I can find say the same thing. NiMh has a high self-discharge. I'll try a NiCd - that should work a lot better.

I have a bunch of NiCds that I got on sale at RS some time ago. They sat in my battery box for years without any care or attention. I discovered a battery charger at a dollar store. I measured the batteries and they all had some charge remaining even though I never charged them when I bought them. I charged them and they all came up with no problems.

So I'm leaning to switch to a 9V NiCd and stop this hassle with NiMh self-discharge.

Mike

That's great, Joseph. I had a lot of problems with eBay when I first started using them, but that was many years ago. Since then, it has turned into a very smooth operation and I've had zero problems. I've only lost one shipment, and the vendor quickly sent another unit.

One thing I stronly recommend is to split your operating system into separate installations. One for normal browsing, and a second installation that is strictly for financial transactions.

I use Ubuntu 10.04 LTS as the host, and VirtualBox as the VM. I have separate installations of WinXP that provide the services needed for browsing, banking, software development, etc. I do pcb layout and routing in Ubuntu so a windows crash will not interrupt an autoroute in progress.

The XP installations are stripped to the bare minimum that is needed for the required services. That means all unnecessary functions and services are stripped or disabled. I turn off Auto Update, Restore, Firewall (not needed), and I don't bother with antivirus programs. Everything else is turned off and any unnecessary files are deleted.

This allows making a very small installation that is easy to back up to a spare hard disk. I use a 10GB dynamic hard disk in VBox. The banking file is only 830MB, and the other installations may be 2GB or so. These copy very fast to the backup drive, so it is convenient to keep the backups current. I can increase the size if needed, and shrink it later when the requirements change.

Then if a virus strikes, or I get a bad software installation that wrecks the registry, I don't have to reinstall XP and all my files, then try to get all the original settings back. That takes a very long time and I never get tham all.

I just copy the backup over the bad file and I am back on the air in a matter of seconds. The other advantage is I can copy the installation to a different computer and get a byte-identical installation on the other computer. XP doesn't know that it is a completely different motherboard and hard drive, so it doesn't care.

I use the System File Checker from Win98 to check for missing or corrupted files, and to see if any new ones have suddenly appeared. This only works with the Win98 version - the XP version does not allow you to specify the file extensions and folders you want to monitor. It also eliminates Win7 since that has folders that will not allow Win98 to enter.

I use the various rootkit detectors to look for keyloggers and other malware. But the combination of Win98 SFC and Rootkit Revealer pretty much covers any trojan or virus that can attack the system, so there is no need to waste time on Symantec or other resource hogs.

The banking installation is the only place where the passwords and logon information to Paypal and the banks is kept. This information is stored in Stickey Password manager and is heavily encrypted. So it is not available to viruses or trojans that may scan the hard disk looking for text strings. But there is little chance they could be on the banking computer since it never goes anywhere else except to the financial sites.

If one of the other installations got infected, there is no way the malware could detect if there is another installation, and none of the keystrokes on the banking computer can be detected on the other installations, so a software keylogger will fail.

The banking computer has no access to email, so a phishing attack on the another computer will fail.

Most of the well-written malware that is aimed at stealing your banking logon information will shut down as soon as they detect they are running in a virtual computer. Virtual installations are used in honeypots to try to find the command servers and lead back to the authors. To protect themselves, they simply shut down and no longer present any danger. So using a virtual installation gives added protection against the most dangerous malware.

Finally, there are two things you can do to vastly increase your protection against online theft. First, open another account at your bank. When you do this, it will have the same profile as your current accounts, and will have the same access rights on a ATM or online. But have the bank add a block to the account, so the only way you can transfer money in or out is by visiting the bank in person and having a manager override the block.

Now you can put most of your money in the account except for what is needed for miscellaneous purchases. This will keep the majority of your funds out of the hands of criminals who would otherwise send it to Russia with love.

The second trick is to change all the answers to the security questions that the bank asks when you set up your online account, such as "What is your favorite color", or "What city were you born in."

Instead of answering these questions with the correct information, give completely nonsense answers such as "!My.Dog.Hass.3.Flees$"

Note the misspelling and punctuation between words. This prevents any phrase search from finding the string.

This is simple enough that you do not have to write it down, so it will not be detected by malware that scans the hard disk looking for text strings.

Use the same answer for all the questions.

Most banks store the browser string from your computer and ask one of the security questions if they detect a change.

Now, if a criminal somehow manages to get your logon information, they will probably be logging on from a different computer and will trigger the security question. But they will not be able to answer it and the attack will fail.

There are many other attacks that can get through all these protections, so you always have to be vigilant and keep monitoring your accounts for unwanted transactions.

But these methods will give a vastly improved security over what you now have.

Mike

OK, here's the original reading on Saturday, Jan 14 2002 at 19:11. It went up to 10.049uF when I took the picture:

Two days later, on Jan 16 2002, at 06:13, it measured 10.045uF with the last digit flickering from 4 to 6. Here it is at 5:

I tried to get a picture showing the reading at 10.046uF, but the best I could do is showing it switching from 6 back to 5:

Another attempt to get it to show 6, but it doesn't stay long enough. It switches back to 5 as soon as I press the shutter:

Occasionally, it may hit 10.044uF, but I'd have to take a movie to show it. That would eat up a huge amount of disk space.

So I'm quite happy with the stability on a quality component. Here's a picture of the capacitor:

I got very similar results on a 25 ohm, 2,000 watt resistor. The reading was quite stable, with the last digit flickering +/- 1.

I should note that a dmm that can measure a resistor to 5 decimal places costs a lot more than the Tonghui.

I think it's a fine instrument and can give an honest 5-digits of resolution. If you see a lot of variation in the readings, I'd check for a poor connection or try a better component.

Mike

I now see the value of an internal battery. It was a whole lot easier to take the Tonghui to the resistor than try to move the resistor to the shelf where the Tonghui was plugged in to the AC adapter.

I can envision lots of situations where the portability would be very convenient, such as working on transmitters, troubleshooting a problem at a customer's site, making measurements on a plane or a boat, and so on. The Tonghui gives the 5 digits of resolution that you normally see on a bench instrument, but you can hold it in your hand and take it with you.

I'm beginning to like this instrument better the more I use it.

Mike

only=20

I hardly ever get phishing email saying it is from ebay or paypal any more, i used to get several per day. Nowdays they all say they are from my ISP.

routing=20

progress.

I like this one, i have been wanting to move my financials to a VM. BTW what PCB software are you using?

=20

=20

=20

I turn off auto update on everything that i can, and a VM is way easy to restore. Have you been able to tell VBox where to put its virtual machines (like to a specific volume)?

a=20

file=20

=20

wrecks=20

=20

=20

I have this one figured out already.

=20

=20

The use of W98 SFC on XP is a new twist for me.

=20

sites.

Cool. Not quite there yet. Still learning to get VMs to do just what i want.

=20

Yep.

=20

try=20

Russia=20

I understand the technique, now i have to reorganize things to make it work for me.

=20

I do a junior grade of this already.

the=20

=20

=20

Except as noted, i do much of this already.

Agreed.

I found one cause of flickering. Getting within a couple of inches of my monitor does it. The display was still readable but flickering in the last two or three digits. If I move the meter a foot or so away, it stops flickering.

The manual mentions something about using it in a relatively quiet environment.

John S

All you need is one that downloads a keylogger. That can come from anywhere.

One thing is to use Gishpuppy for disposable emails. It works great in Firefox, and you can have an unlimited number of email addresses.

I was using PCAD but it has outlived its usefulness. However, it had macros, which means you can do wonderful things with your own code.

For example, I could overlay a previous layout on the current pcb and show exactly where the design had changed. I also made my own parts selection menu that held my complete inventory, plus automatic links to datasheets and any other useful information.

I'm looking at Eagle since I understand it also has macros. I'm hoping to re-use most of the code and simply convert the commands to the new syntax.

Do you mean to run the vm on a flash drive? No, I'm not good enough to figure out how to do that, but my experience with Ubuntu says it should be possible to do pretty much anything you want. All you have to do is find someone who has done it and posted the procedure.

One thing I forgot to mention is to turn off the Windows cache. This simply wastes disk space and has no benefit.

Another thing I should mention is when you are generating the vm, don't use the latest release of VirtualBox. It will put the vdi file in its own folder and separated from the other vms, which makes it difficult to back up.

Instead, use version 3.2.12 to install the vm. This will put all your vdi files in a single directory where it is easy to pick which ones you need to back up and do them all at the same time.

However, version 3.2.12 takes forever to boot the vm, so you want to go to a later version which fixes the slow bootup.

However, don't use the latest release to run the vm. This has another annoying bug.

VBox added support for Windows Aero starting in version 4.1. Here's the release note:

"July 19, 2011. Oracle today released VirtualBox 4.1, a new major release. Introducing VM clones, increased capability for very large hosts, and Windows Aero support."

This causes the probe in LTspice to align the tail to the wire instead of the pointer. The problem is described in the LTspice forum

I found that all the version 4.1.XX releases do the same thing. But if you go back to version 4.0.12, it fixes the problem. This is the last version that does not have Aero support, and was released 4 days before. Here's the release note:

"July 15, 2011. Oracle today released VirtualBox 4.0.12, a maintenance release of VirtualBox 4.0. It improves stability and fixes regressions."

Also remember to update the Guest Additions when you run 4.0.12.

It is available online. If you can't find it, I will dig up the url.

It is a marvellous little program that can detect a single bit change in any file. So you can see if a virus has changed any of the Windows system files or downloaded other files that you don't know about.

The malware can change the operating system to make the changes transparent. But if it does this, Rootkit Revealer will show the infection.

So you get them coming and going. If they don't mask themself, Win98 SFC will catch them. If they do mask themself, Rootkit Revealer will find them.

If you figure out any useful tricks, please post!!!

I think these methods will block most malware attacks. It used to worry me a great deal that I would lose everything in my bank account and credit cards, since there is almost no way to avoid doing financial transactions online.

But now that I have blocked the bank accounts and split the vms, I can do all my banking and Paypal transactions and I am now free of the worries that some rootkit or trojan can get to the account.

I also use many other methods, such as Proxomitron to filter IFRAMES, Drop My Rights to prevent Firefox from installing files in critical system folders, QuickJava for quick enable and disable of Java, Javascript, Flash, Silverlight, Images, Stylesheets and Proxy from the Statusbar, and so on.

I used to use Certificate Patrol to check for changes in the SSL certificate which would indicate a malware had changed the dns to point to a copy of the banking site. But it turns out that banks may change their certificate often enough that it just became a pain to try to keep track.

Finally, the Win98 SFC method won't work on Win7, so I am committed to XP for the forseeable future. But I have no interest in Win7. The basic install requires 6GB since they have included every possible variation of every DLL they have ever released, and put it all in the WINSXS folder. So there is no way to get rid of the unwanted junk that you will probably never use, and the vdi now takes a very long time to back up. Typical Microsoft crap.

I'm happy with XP. It works fine, and I don't need the fancy resource hogging features in Win7 and later releases.

Thanks,

Mike

That's very interesting! Thanks. That means to be careful of switching power supplies and SCR dimmers.

Are you using a CRT or LCD monitor?

Mike

LCD.

Hmm... I just repeated the experiment and I don't see much change. I'll have to be more careful about my observations in the future.

Sorry!

John S

Don't be sorry. You saw the readings fluctuate. That's exactly the kind of thing we are looking for. Some erratic performance, or a funny glitch, or strange behavior. I did the same thing when I first got the unit and tried to figure out how to get it to work with a dead battery. My observations and conclusions were way off the mark.

Now I wonder if perhaps the Kelvin clips weren't making good contact, and moving the unit away from the monitor broke through the oxides and allowed better contact?

I think maybe we are used to 3 1/2 or 4 digit dmms. So we are not sensitive to making measurements with a 5 digit resolution, especially on components that have been sitting in a junkbox for several years.

So far all the quirks have turned out to be violations of the operating notes, poor quality electrolytics, or perhaps poor contact on the connections. But we need to find these things and list them so we can get the best performance from the unit.

So don't be afraid to post any anomalies that occur. I think it was Bob Pease who made sure his support personnel understood the need to question any strange behavior in the circuits. Good advice.

Thanks,

Mike

All excellent points. But, I need to verify my observations before posting.

If you have a 60Hz transformer around and you have some time, please make some measurements on its primary. I'm finding that a couple I have are showing in the half-henry range, yet current/voltage measurements on the line are showing about 10 times higher. I'm wondering how to track down the discrepancy.

Thanks, John S

I'd still be interested in any quirks you find. These may occur only rarely, and be so fleeting that it's hard to figure out what is happening. But if we can collect enough observations, maybe we can pin it down to something that is easy to prevent.

I'm not sure what you mean. Is the unloaded current much higher when you apply 120VAC than the inductance measurement says it should be?

I wonder if there's an issue with waveform distortion or some other problem with the current meter. Most handhelds do not measure true rms, so if the transformer is saturating or there is distortion in the line waveform, they may not give a correct reading.

But I am interested in following up. I just bought two handheld current clampmeters on eBay, and I made a simple junction box with a small resistor in the neutral lead so I can look at the current waveform. So let me know more about the problem and I'll be happy to take a look.

Mike