Would you buy a MAX1668 based Temp monitor(kit)?

[MMZ_TimeLord]

Almost got the first one redesigned... and I got the recommended schematic from Maxim for the 6655 chip.

Should not be a problem to put two of those on there.

Boards are now exactly 1" x 1" AND I added the ground pins, the power switch and TWO SMBus+power headers to daisy chain.

The power switch will only turn off the immediate board. Power will still be provided to the chain.

Shoud be finished tomorrow.

[Cptn. Foo Foo]

Originally Posted by MMZ_TimeLord

The power switch will only turn off the immediate board.

What kind of switch? Micro surface mount? Just wondering.

I am also more interested in the higher accuracy version BTW.

[pHaestus]

Any thought to how you're gonna get all the extra bits? MBM will only report w/ 1C resolution unless you can manage to slow the sampling rate. You may need some extra layer of software to send commands to ICs...

[pelle76]

Heres my implementation of the 1668. Works a charm!

I hooked up the SMBus on the backside of the mobo, on the RAM socket solder points. I also made a connection on a RAMstick... But the backside of the mobo was far more easy to solder and that is tha connection I use





EDIT: The Mobo is an Epox 8rda+ but any mobo with DDR should have the same pin configuration

[leejsmith]

i tried to make my own but made a right mess of it.

would you ship to the uk ?

[murray13]

Originally Posted by pHaestus

Any thought to how you're gonna get all the extra bits? MBM will only report w/ 1C resolution unless you can manage to slow the sampling rate. You may need some extra layer of software to send commands to ICs...

Have not looked into it fully but I do know that you can set up MBM to send commands over the SMbus. :shrug: It's in the help file somewhere.

[BillA]

nicozeg
sorry, just chasing a good idea (also)

[MMZ_TimeLord]

Murray13,

Thanks for the cable reference... I'll see if I can get that.

Capt._Foo_Foo,

The switch is a micro toggle PCB switch SPDT. Digikey part # 360-1012-ND

pHaestus,

MBM5 I believe supports these as most motherboards already have at least one similar to the 6655 onboard. The madhacker article I found though the MBM site under the "Extensions" tab.

pelle76,

Did you realize you have a spider in the middle of your circuit board?

leejsmith,

Outside the 48 states, the purchaser will be responsible for shipping. Shouldn't be much as small as this is.

[pHaestus]

yea it detects fine with mbm. will report temps with 1c res though. ic capable of 1/8c; just need to lower sampling rate and reclaim those bits...

//edit (underlined)

[pelle76]

Originally Posted by MMZ_TimeLord

Murray13,
pelle76,

Did you realize you have a spider in the middle of your circuit board?

Yeah... I was a little suprised that the thing worked... ...I mean, thats a quite big "BUG".

[MMZ_TimeLord]

Here's the new design fellas... The SMBus & Power will use a 6 pin connector. I suppose I could have made it five pin... but I wanted a key.

Let me know if this is to your liking... working on the 5566 now.

[pHaestus]

MMZ did you check the Maxim guides to trace width? There's a pdf about improving accuracy w/ diode-based solutions on their site and they give some guidelines. With length so short (and with header there anyway) it shouldn't be a big deal but might be worth perusal. That 0.125C bit is a big deal though for some. I bought $100 MAX6655EVSYS purely because they had software that did 1/8C

And of course that software only works over parallel port using their SMBus to Parallel adapter...

[pHaestus]

Oh and I'd be good for 1 of each (soldered).

[nicozeg]

Originally Posted by Groth

The catch is, it isn't a differential measurement between the Vcc and Vcc pair. It's a single measure relative the the IC's ground. If you're using separate computers for your taget system and for your measuring/logging, you'll have problems/errors.

Could that be solved by joining grounds of both PCs?

In fact that happens through the power cable to the wall, but maybe a more direct path can be better.

Originally Posted by pHaestus

yea it detects fine with mbm. will report temps with 1c res though. ic capable of 1/8c; just need to lower sampling rate and reclaim those bits...

//edit (underlined)

Why stick to MBM if we can use Maxim software with the full resolution available

As I understand, your compatibility problems with win XP were more related to the paralell port interface used in the evsys.

[nicozeg]

Originally Posted by pHaestus

And of course that software only works over parallel port using their SMBus to Parallel adapter...

Ouch I did't know that, That software is suposed to work with the Evkit which is the smbus only version of the Evsys.

Have you tried to use it directly attached to the smbus mobo header?

[pHaestus]

Yea see where it says MaxSMBus? That's their daughtercard... I'd be happy if it DID work though (no mobos with SMBus headers onboard atm can someone else check?)

[pHaestus]

The same PSU usually powers the MAX6655 and the test PC. That was the source of my weird data a couple weeks ago; PSU was going bad and was only supplying something like 4.4V to the MAX6655. Giving it juice within spec returned results to what previous testing had shown.

[BillA]

sounds like a good argument for NOT using the computer psu (in any case if the supply voltage is affecting the output)
pHaestus - any idea of the sensitivity ?
(is a STIFF 5V psu really necessary ?)

[pHaestus]

.pdf says 4.5-5.5 for VCC; no data for effect of VCC on temperature in specs but every other parameter is graphed.

I had some serious problems with that PSU though and AMD CPUs will stress the 5V rail. I guess a PC PSU that allows you to dial up the 5V (as measured by Vin on Maxim) would be the way to go if you wanted to monitor VCore at the same time. Thermaltaco sells some with potentiometers in a bay drive for adjusting power. Alas! If I only had a grothmeter (and a way to convert output current to a voltage I could read with maxim Vin...

[MMZ_TimeLord]

Okay guyz... I think I got the dual 6655 done... WHEW!

Man that's a crowded board! Same basic layout as the other one with the temp probes in the same place with the same pin layout. The SMBus header and power headers are in the same place for consistancy.

I did include pins for the voltage probes and the additional address selectors. There is also a two pin header by itself for each of the OVERT pins (both OVERT pins in one two-pin header)

I'll have to make a sticker or something to mark the back of the board as there is no room for silkscreen work.



Let me know if you spot any obvious flaws before I go to prototyping. (probably next weekend)

[pHaestus]

Ok scratch my order for one of the 1668s; would like 2x these. How much different in price for dual boards?

[MMZ_TimeLord]

About $10 more each for the parts cost increase the overall soldering work is about the same.

[pHaestus]

So looking at about $50 now? I will probably back down to one and save the money for other activities now that I think about it.

[MMZ_TimeLord]

I haven't actually run the numbers pHaestus.... lemme narrow it down. I MAY be able to cut it lower. But I need to get the actual prices from DigiKey.

I'll let y'all know.

[Groth]

Originally Posted by nicozeg

Could that be solved by joining grounds of both PCs?

In fact that happens through the power cable to the wall, but maybe a more direct path can be better.

Problem there is ground float. The return currents times the wiring resistance means that 'ground' at the little IC board will not be the same as 'ground' at the CPU socket of the target PC. You could try to tie the ground of the two PC together, but you have the risk that some of the current supplied by PSU-A will want to return via PSU-B, and switcher power supplies don't like that.

In my experiments, I used an optically isolated SMBus. The IC board was powered by the target PC and grounded at the socket, but the control and logging was done on the electrically isolated second PC.

Perhaps our good buddy MMZ_TimeLord could make a small, separate SMBus isolation board?

pH, the Grothometer is pretty much useless until/unless I get an o-scope to calibrate it. And with the ongoing underemployment, that might be a while.