EEVblog
ฝัง
- เผยแพร่เมื่อ 2 มิ.ย. 2024
- What is the main chipset used in the EEVblog BM235 Multimeter?
Dave finds out.
www.ic-fortune.com/
www.cyrustek.com.tw/
www.hycontek.com/
www.hycontek.com/attachments/D...
Forum: www.eevblog.com/forum/blog/eev...
EEVblog Main Web Site: www.eevblog.com
The 2nd EEVblog Channel: / eevblog2
Support the EEVblog through Patreon!
/ eevblog
EEVblog Amazon Store (Dave gets a cut):
astore.amazon.com/eevblogstore-20
T-Shirts: teespring.com/stores/eevblog
💗 Likecoin - Coins for Likes: likecoin.pro/@eevblog/dil9/hcq3 - วิทยาศาสตร์และเทคโนโลยี
I once had to repair a TV for someone. And on the main board, a 0 ohm resistor went open, and caused a quirky issue of where it's service settings kept getting loss. I honestly had no proper 0 ohm resistor jumper, so I just took a piece of wire, and was going to jumper the pads. The guy was going mad at me, said I was "hacking" his TV "beyond" specification. I tried to explain 0 ohm resistors are just doing the same thing I was doing, and it's only cosmetic as during manufacturing, it's easier to use a 0 ohm resistor than a piece of wire. Oh man was he not having any of that!! Thanks Dave, when you mentioned about 0 ohm jumper resistors, you just brought back some really bad memories.
That was back in 2008. I learned after that, just don't tell the customer what the hell you're doing, leave them in the dust. It sounds rude, but it's a lot easier.
+killaken2000 But it would be no different. A piece of wire has no resistance, just like a 0 ohm resistor, and a cut down tiny piece of wire would be just as clean too, and cheaper, as I don't have to buy a stupid 0 ohm resistor jumper.
Wire + Stimorol + black marker = 0 Ohm resistor for your customer.
I honestly can't understand your logic. It's the same exact repair. It's going to be a repair in any event, and there is no significance to the choice beyond part sourcing and manufacture. A 0 ohm resistor *is* a jumper.
How about I just jump a wire where a 0 ohm resistor should be and charge you as if I replaced a resistor?
I've never even charged for resistors. Caps, yes, because I use only name brand caps like Nichicon, UCC, or Panasonic, but resistors I never charged. I still charged for labor though.
please more Reverse Engineering! I love to see how products are designed at the lower lever and the challenge on figuring out protocol and innovative solutions to real world constraints.
I always make sure to get 1% 0 ohm resistors, the 5% ones just aren't accurate enough.
Yeah, trap for young players.
This just made me snort out beer out my nose. Thanks!
This is really important. Otherwise you could end up with a negative resistance as these tolerances work both ways!
Good thing I spotted this before placing a order for 6.3M 1608 0ohm resistors, going to make a 10km 0ohm conductor, that 5% could've ruined it.
Nope I always pay extra for 0,01% 0R resistor for superior performace
The EEPROM won't just die when it hits 1,000,000 writes. That write count is for each byte. The bytes that have less writes will still work fine. Even after exceeding the million writes, the byte still works, the reliability just starts decreasing.
Dave, I am so addicted to your videos that during the last 2 days I was feeling depressed and didn't know why until I came to watch you! :) Love you man ! :)
Just wanted to comment my assumption that one of the dual EEPROMs is probably physically write-protected, but you figured it out yourself. A question about star grounds and power routing: how viable it really is? Is it true that they can pick up EMI due to long individual traces? How does star compare to filled ground plane? A video on correct analog ground routing would be great!
Noticed the Louis Rossmann-like format! Love seeing you guys exchange ideas.
Err, no, I've been doing screen cap vids for most of the EEVblog's life.
+EEVblog No worries Dave, the vast majority of us have known for years now who's the student and who's the master...
@@EEVblog I think that guys "Rossmann" comment was a bit cuntish.
Hi, Dave, one piece of evidence you may have missed is comparing the plastic cases of main chip and the LCD driver chip. Different manufacturers will have visually different cases, but these two look absolutely identical. Finish of the top surface, the location, size and shape of the pin1 marker, the rounding of the edges, shape of pins, etc. It all points to the same manufacturer of the cases. Of course this is not 100% proof - the chips could still be coming from different fabless companies that happen to use the same manufacturer of chips. One more bonus hint: both chips have H200 written at the same spot.
I noticed the H200 as well.
As an embedded software engineer, if I knew I had an entire 2k EEPROM to work with and had to update a setting very regularly (like once per second for a clock or every time a mode change happens), I would split the data up into a linked list of records and write them into the EEPROM successively. With a EEPROM you can usually change a bit from 1 to 0 for free and you only impact the cycle count when you erase a sector back to all 1s. So you can have a "validity" bit in each record that you set to 0 once you have written the next record successfully and this will tell you when you read back the data to ignore that record and move on to the next. Then you would only erase the part once you have filled up the entire 2k bytes with records. By doing this your 100000 or 1 million erases will last you years and years even if you write to the part once per second.
For example, if you had 4 bytes of mode information to store, you can make a 5 byte record (or put an index table at the beginning of the part) and get roughly 360 records in 2k EEPROM. Then, if you write a new record once per second, the part will still last for 10 years instead of just 11 days.
Dave, if you are still reading these comments it would be interesting to see a video about reverse-engineering the data format of the EEPROM to see what information it stores and how it does it. You could show a logic analyzer like the Saleae in action or examine the dump from a EEPROM programmer.
Thanks Dave, very interesting. It's great to see a manufacturers website with such comprehensive information on their products and available to all.
Bonus for not having to register with email in order to download even the most superficial datasheets, let alone internal details such as the instruction set of the chip's processor. I've never seen any of the big ic manufacturers do that, which is why I have to keep using fake accounts for the registration and/or downloading datasheets from suspicious sites.
Great stuff! Getting the itch to rip out the Saleae Logic Analyzer.
Highly recommend the BM235. We use it every day.
probably got rid of the LCD capability to get the extra 1000 counts?
I also was thinking about it.
I love the auot range ICs. they're much better than the auto ranging ones.
Loved the Ghostbusters-esque reference ... Thanks Egon!
Not the first time in my videos!
+EEVblog Nice to see that people still appreciate the classics ... after watching the new movie, more than ever!
Perhaps the don't use the internal LCD driver because they wanted to use the flash memory space to provide more functionality rather than for LCD logic.
Dave,
GREAT BLOG.........How about a class on Volt Amps. Real power vs Apparent power. There seems to always be confusion on this subject. Thanks.
George.........
Finally a good video again Dave :) The last 6 or so have been a kangaroo looking for a rabbit.
15:54 Is Dave working on a meter design?
By some weird coincidence, I'm trying to repair one instrument with an HY3131 chipset inside... It's a 50.000 counts current clamp meter where they didn't bother to remove or alter the markings.
Hi Dave,
I noticed on the picture of the BM that the external oscilator pins are 14 and 15, and the datasheet of the micro you were considering to be the on on the BM says pins 15 and 16 for the external oscilator.
Great video!
Very nice, thanks!
Great Digital Mult iMeter ! The only con is that it is not a Fluke. Bought this on one for the up-graded leads. Thank you for the video ! Had a little bit of HF electronics in the Army... tjl
BTW...Waffle on as much as you like Dave...After all it's your channel! :-)
If I would like to extend the life of the EEPROM, I would set a circular buffer to let the firmware write data in different location each time the meter is powered on. This is possible because it needs a couple of bytes or so for mode saving, and the EEPROM has plenty space for that.
Was thinking exactly the same thing. But how would it know where to read/write the next byte? Do these things have working RAM?
Would it require more than 1 byte for mode saving? 256 different modes can be stored in a single byte. I guess we'll find out in the next video...
+Prehistoricman. In my option, I'll let the firmware do a quick search for a certain byte, let say 0xAA which will be followed by the last save location.
Mohammed Algailani Then you're wearing out a certain cell of memory storing where your other bit of memory is.
+Prehistoricman. The 0xAA byte is written just before the new data, not stationary in a fixed place. After powering ON, firmware search for 0xAA, read data then erase the 0xAA so that the next power on there is only one 0xAA.
Mohammed Algailani Ah yes! I get you now. I had that idea too but wondered how fast searching for that cell might be. I guess for initial turn-on, it would be acceptable to wait 0.1s while it finds that 0xAA.
Going more in depth on this meter would be fascinating (not just saying this because I own one!). Is there any data on the unused LCD pins? If the LCD output section is disabled as an internal option, maybe it runs a bit faster?
not a clue, but the lcd uses multiplexing, so maybe the lcd coms from the main chips are used for something else(maybe buzzer?) or maybe to make the layout neater, to reduce the number of layers
I think you're right about segmenting the EEPROMs by purpose. Also, it's probably more efficient to stock the same part for both purposes regardless of how much storage is actually needed. The 2402 is a 2k _bit_ EEPROM, or 256 bytes. The main processor doesn't even have enough RAM to buffer all that data.
Yeah, it would be BOM reuse once they decided they wanted them separate.
It's very common to have ROMs bigger than the microcontroller RAM. In most cases, you really don't need to load all the data into RAM. ROMs are random access for read
Did you meter the supposed supply rails of the chips to see if it was the supply pin as indicated and it was in the spec range? ok so some other issues
There is a way to not wear out an EEPROM. You can write in cycle to the next adress every time. Microchip has a library to do this - DEE Emulation.
Dave I think this further EEPROM investigation may be very interesting, so it maybe shouldn't go into the second channel.
If I were doing the firmware for that, I wouldn't save off the mode until it had been in that mode for 5 seconds. No point writing out every mode you click past. If they really meant that mode, they'll be in it for 5 seconds. Then even if you're changing modes exactly once per 5 seconds constantly, it would take 1389 hours to wear out the flash chip. 57 days of constantly changing modes every 5 seconds - or doing it 8 hours a day for half a year before the chip wears out.
And that's a minimum spec, odds are it'll last twice what the data sheet minimum spec says.
You could probably find out if they have a delay by probing the I2C bus on the chip then changing ranges. EDIT Oops you said you'll do that. Sorry, typing too fast.
balbes1234 Are you not then just wearing down the location that tells you where the current write location is? (W) I can think of ways around this but it could easily just transfer the problem to a different location.
balbes1234 yes of course but I was wondering how you know which of the pre-allocated values is actually the current one? Do you just preload addresses with a known invalid value (0, or 0xFFFF or whatever), then read through all the values in order until you find the last valid one to know which one is valid? Then when you fill them all up, do you go back through and erase the whole stack to start over again? That was one of the methods I thought of.
Is the boardlayout of the bm257s similar to the bm257(without the s) or are there differences?
When I see those two "L02" SO-8 packages you are pondering about hanging like that I was screaming inside "That's gotta be AT24C02." Those I2C EEPROMs all have the same pinout and protocol (alright, two different, mutually incompatible protocols, but still compatible within its own class,) and are largely interchangeable. So maybe it is a ROHM part after all but to me those are unbranded AT24C02s.
Yes, could be Atmel or Rohm parts. BR24L02 from Rohm is only 0.20$ in quantity so they are not really spending that much more for the second one.
Christopher Johnson Rohm BR24L02 or Atmel AT24C02 or Microchip 24AA02 those I2C EEPROMs are all compatible with each other.
27:00 that crystal looks like it is connected to pin 14 and 15 unless they really connected it trhough that via leading to 16. It's off by one pin. I guess the marking "17" on the silkscreen has done it's deceiving purpose :) Only one way to find out. Beepy time :)
Yep, I checked and it is 14 & 15.
Hmm intersting. I still think you are right about it being a Hycon part customized for Brymen, and this is just differnt bonding. Looking at bare die pictures they almost always have excess pads. If that's the case for the Hycon chip they could create a "custom" Brymen chip by just bonding out those prior unused pads. That could explain the different function on the LCD pins and the offset crystal pins.
I suppose you could check if it is writing each time by changing something and then pulling the batteries rather than powering it off normally. Then putting the batteries back in. I suppose they could have the caps hold a small charge to make sure the write occurred if the batteries were pulled but it doesn't seem worth the effort.
You forgot to compress the audio to -0.1DB.
I didn't bother, just up the volume a bit, it's consistent.
It is already limiting out the usb sound card with the 50mm drivers.
Download the video with youtube-dl, and amplify it with VLC or any other media player
Then I'd say something is wrong with your setup. I'm able to play it at deafing volume.
Audio is absolutely fine
I've been looking at some multimeters online today including here on EEVblog...I think I'm leaning toward a Gossen.
They look so solid. Quite apart from the fact that German built stuff is great, they do seem to offer some potentially good value models. It all depends of course what you envision using it for.
What's expensive on the Gossen ones are their optionals.
The HY12P65/66 datasheet says crystal is on pin 16/17, but looks like it's 15/16 on the board.
No! Look closer, crystal is connected to pins 14-15, not 15-16 as per DS. That's the first thing I noticed and was ready to shout about t, but decided to watch if you catch it. You didn't. :(
Ah, I got confused by the pin 17 placement, doh.
That was the first thing I noticed as well as many other viewers I am sure. BTW, awesome rev-eng content, it is very informative I think, big thumbs up. I hope You'll continue such practice in the future.
IRFANVIEW-> my favourite!
I did wonder if that J1 was a cal jumper, being close to what looked like EEPROMs.
I got a free loose jumper in my BM235 (presumably formerly positioned on J1), so had to follow the "Don't turn it on, tear it apart" motto. (Murphy is always waiting for you to power up something with a loose screw or other metal component strategically located to cause maximum damage.)
Left inside by the cal tech! LOL
+EEVblog Yeah, probably dropped it, and usually under significant time pressure so can't spend all day trying to find the wee bugger.
There's no point to keep that secret from Dave Jones , He always know it.
Probably had to disable the LCD interface to get the back some Flash space (if the LCD controller is just I/O and it is implemented via the microcontroller) so they would have room for your requested features. :)
LCD controllers usually have specific driver hardware on the pins.
They made the LCD driver chip design decision before I spoke to them.
My thoughts exactly.
Removing the LCD controller could be a way to reduce its noise leaking to ADC. LCDs need to be constantly regenerated via specific periodic signal patters on multiple pins thereby creating noise on the chip.
You would expect that the chip was designed with that understanding and that it accounts for that. Then again, perhaps that is why the chip claims 5000 count performance. Perhaps BM is able to squeeze the 6000 counts out of it by reducing the over sampling required to account for the noise the LCD introduces...
Maybe they can also use the extra eeprom to store all the extra firmware code that were needed to satisfy you :)
The soldering on the Brymen 257 is terrible. Plus, is there a part soldered over some passives on the lower right of the board?
To me the soldering doesn't look that bad. And I think you mean the through-hole plastic film capacitor, which is bent over some passives, so that it fits under the LCD. That cap is probably for the ADC, which requires a linear, low leakage capacitor, hence the plastic film cap.
volume is a little low
What about intersil? The first thing came to my mind, but didn't make its way to Dave's top list...
Cupon already expired? Better be quick here at the EEVblog, 3 days!
do a part 2! :)
Hi Dave do you have test equipment to give it away
Thanks
They may have some extra firmware code in that I2C flash that is programmatically loaded when needed.
That was my first thought as well. Even though it was an OTP part, they could patch the code. If so, they could apply a firmware overlay during calibration to patch the OTP code. Some chips like the OTP Intel 8051 processors can be forced to boot from an external EPROM which makes them still useful pulled parts for playing with.
What are 0-Ohm resistors useful for? Is it to jump over traces?
Yes, plus breaking into the circuit during design debug, plus design options etc. Plenty of uses.
Makes sense. Thanks, Dave.
A part two!
How did you identify those three vendors as the main suppliers of meter ascis? Just tribal knowledge or did you apply a methodology?
You can identify the makers of microcontrollers for multimeters via an internet search. Maybe start with multimeter microcontroller. One rather important maker that Dave didn't mention is Texas Instruments. Here's a recent (June 2016) Texas Instruments design using its MSP430F6736: www.ti.com/lit/ug/tidubm4/tidubm4.pdf
As I understand it, Texas Instruments makes the processors in high-end Flukes (don't know if it makes them for the Flukes made in China).
what is ment by "8x8 hardware multiplyer"? what dose a Hardware Multiplyer do?Also why is this desired?
Multiplies two 8-bit numbers. Doing it in hardware is much faster than doing repeated addition in a loop in software. en.wikipedia.org/wiki/Binary_multiplier
So with coupon one could get EEVblog marked Brymen below your costs (AFAIK you sell them with almost no profit, am I right?)
already got one anyway, happy to use it.
Good Waffle can go on :)
Perhaps they needed more I/O off the main cpu then used the lcd device to free up pins.
They probably have two EEPROMs so that they can write protect the calibration.
EDIT: I guess I should have watched the whole video. (37:20)
video on hacking the i2c bus would be cool.
I need one.
What are the chances that they got a deal on ICs with bad lcd controllers and had to use an external one?
Unlikely you'd get bad dies in just one area.
Maybe using the onboard controller somehow generates too much noise to get 6000 counts, or maybe they have to borrow clock ticks to run the LCD driver, but by eliminating it, they have an extra few cycles to sample with.
Doh, if I only could have waited 3 days before ordering...
WAFFLE but i bought mine months ago. Another interesting video
Hi Dave!... I smell something fishy in the air... could it be.... power from footsteps?
Let's tap the i2c!
Decap the IC and throw it under the old microscope!
2 eproms raid 1 mirroring
Dave, I think you often use 'chipset' incorrectly.
Yup, I also pointed this in one of previous videos.
Meh.
EEVblog After your rant about the pronunciation of [geo]cache, one is left with the impression you care about such accuracies.
cheap