treeman wrote on 2023-03-01, 10:17:

On my pcb I have a standard 14.318mhz crystal next to a chrontel clock generator. Using a multimeter in hz mode I got nothing direct from the crystal, got 6mhz for a second and seen it once or twice during many tests. So I had a spare crystal swapped it and still the same.

Different chips do different things. In this case the chip most likely has a resonator/amp circuit, a PLL (or several) and some divider ladder(s) for the outputs.
You can't just shove meter probes into modern low-power crystal oscillator circuit and expect it to still work. In many cases the meter is a "perfect" device that draws no current, has infinite imput impedance and no parasitic properties like capacitance (input and mutual between leads), etc. But not in this case. You are loading it down and it stops working - you need a scope with 10:1 probes to measure these things properly. That being said, in a circuit like that one of the pins is input, one is output. A good meter might just be able to measure the ouput (to GND) without loading it too much, but still a scope is much better tool for this.

You should see some of the chip pins output some frequencies. There might be ISA clock, KBC clock (if it's a separate chip), main CPU clock (this will depend on jumper straps), and all these are outputs with decent drive. If you can't measure any clock activity from the chip, it's either dead, has no power, or there is something wrong with the oscillator circuit. A short, or one of the capacitors you mentioned is open for example.

The datasheet usually gives enough information to decode what is going on. On many older boards the 14.318 Mhz crystal is used as a reference to the clock chip which generates same plus 8Mhz for ISA, the CPU clocks and Peripheral, some have PCI or VESA clocks.
example of two PC clock gens: Chrontel CH9001: https://datasheet.datasheetarchive.com/origin … A2IH0035467.pdf
and Macronix MX8315PC: https://datasheet.datasheetarchive.com/origin … AIH00089031.pdf

Thank you for the explanation. I thought the function is something how you explained it. The real work is done inside the clock gen chip using the initial signal from the crystal.

As for the meter I was sceptical a multimeter can do the job of a osciloscope so now I know I am still working half blind.

As I only have a multimeter what I am able to test is:

The crystal is getting roughly 2.5V on both legs, I think this is correct.

On pin 9 of the clock gen I actually see a 14.318 signal but nothing on any other legs. Like you say maybe it is because my meter just can't see it.

The Chip that is used is a chrontel ch9007D-N I can't find any data sheet or good information about it.

Since I changed the 14 mhz crystal and both behave in the same way I think I can deduct it is ok.

The capacitors are not shorted and they send around 2.5V to each leg of the crystal, I think this is normal and ok too.

It leaves me with the actual clock gen chip, it is a hard chip to source so at the moment I don't think I can do anything else.

Tiny drop of high pressure blinker fluid inside which can evaporate over time...

Just kidding, above look like they're steering you straight, if you want ground up first principles search Phase Locked Loops and Frequency Multipliers and Dividers.

You don´t see Octec VLB mainboards too often.

I guess - looking at the footprints -if you can´t find the Chrontel chip, you can remove the X-Tal and the Chrontel and replace them by a 4-leg standard oscillator (25 MHz or 33 MHz) that has both X-Tal and oscillator círcuit integrated.
You can´t jumper for 25/33 MHz any longer in this case, but if you put the new oscillator into a socket, you can change the oscillator if needed.

Ther might be some changes of the "jumper wire" settings nearby needed.

treeman wrote on 2023-03-01, 10:17:

I have a 486 motherboard which I suspect the clock generator is not working.

Sorry for being disrespectful, but someone who doesnt know how to measure crystals and has no tools able to verify clock output is not the best person to make those guesses.
Start from basics. what board is this? https://theretroweb.com/motherboards/s/octek- … -1-486-rev.-1.0 ? https://theretroweb.com/motherboards/s/octek-hippo-vl-2 ?
describe symptoms
Plug POST card and tell us what it displays.

The board is octek hippo vl+ 3.02

I had this board for a few years, I try a few things on it once in a while. The board never showed signs of life. No beeps no signal, the voltage is present in the rails, chipset bios etc. I even recapped it about a year ago.

I also don't have a post card, I work with less tools but as hobby it really doesn't bother me.

I think clock signal, voltage on cpu, reset signal are all basic things to look at first so this is what I am doing.

Using a 4 pin crystal instead of the clock gen a 2 pin oscillator sounds like not a bad idea, but without a data sheet for the clock gen im not sure how to wire it up

what cpu, how are jumpers set?
picture of the board?
multimeter model? what frequencies can it measure?
POST card is $7 including shipping nowadays ("4-Digit Card PC Analyzer Diagnostic Motherboard POST Tester Computer PC PCI ISA")

I have a vici vc97a
Tried dx2 and dx cpus at 25 and 33fsb

I tried the clock pin on the cpu slot and it doesn't show any hz, but in volt mode it shows 2.5V ~

Do you think it doesn't pick up the mhz in the cpu slot?

When I try the 32mhz crystal it shows 32mhz ok

"Frequency: 10Hz/100Hz/1KHz/10KHz/100KHz/1MHz/30MHz"

so it should show at least in 25MHz jumper mode

>I tried the clock pin on the cpu slot and it doesn't show any hz, but in volt mode it shows 2.5V

with and without cpu?

measure frequency and dc voltage on every pin of Chrontel chip, there should be 14MHz (buffered output, not the raw crystal) and 24/33 somewhere there. Most likely one or both of those clocks go to 74fct244 chip next to it for additional buffering and redistribution.

>I tried the clock pin on the cpu slot and it doesn't show any hz, but in volt mode it shows 2.5V

2.5V does look like there is clock running there, and your multimeter fails to pick it up. Maybe, just maybe one of the frequency jumpers has bad contact and instead of 1-2 1-2 2-3 2-3 you end up with some other combination sending 50 or 66MHz, that would not work on old 5V DX/DX2

I found the Chrontel CH9007 datasheet but only has ver A and C, the 8 pin versions. On a chinese website found reference to the 14pin D variant like on Octek board and a reference to ICS AV9107 as a replacement.
Comparing the datasheets looks like the 8pin version of the ICS 9701C-05 is a replacement for the CH9007A 8 pin, and the ICS 9701C-10 is for the and CH9007C.
That being said it is very possible that the ICS 9701C-03 or -11 14 pin could be replacements for CH9007D...
Someone want to double check that ?? Attached both datasheets....
added: there is an error in the PDF of ICS, one of the tables is bad for Table for AV9107C-03...found a better pdf.
labeled ICS AV9107C clock gen 1995.pdf below (replaced the other one)....

I measured with cpu and without and there is no HZ display.

On the chrontel chip only on pin 13 I see 14mhz / 2.5V like you say. Other pins of chrontel show no hz but a few show 5v and 2 pins show 2.5V.

74fct chip shows no hz on any pin but shows voltage on many of the pins to be 2.6V

I checked clock jumper pins and they show continuity with the jumper on, they also show continuity to some pins on the chrontel chip.

Can you check in Ohms (with computer off) if Jp5 middle pin connects to Chrontel pin2 and find which Chrontel pin is connected to middle pin of JP4 ?
Also check which pins are near zero ohm to +5 and ground. Pull the jumpers first 😀
It may not help you but might help figure out the pinout of the Chrontel....

Ok I poked around the pins. Looking at it from the angle I drew the picture.
The 2 middle pins only have 1 connection each pin 1 and pin 2 on the chrontel chip

the other 4 pins top 2 and bottom 2 are all connected identically to the chip pin 3 4 5 9 10 12 14
pin 13 gives me 14mhz which doesn't have a direct connection to any pin
pin 7 and pin 8 connect to the crystal pins

my drawing is confusing but if you consider that top 2 and bottom 2 are all the same you see a picture where they all go

Horun wrote on 2023-03-03, 00:26:

I found the Chrontel CH9007 datasheet but only has ver A and C, the 8 pin versions. On a chinese website found reference to the 14pin D variant like on Octek board and a reference to ICS AV9107 as a replacement.
Comparing the datasheets looks like the 8pin version of the ICS 9701C-05 is a replacement for the CH9007A 8 pin, and the ICS 9701C-10 is for the and CH9007C.
That being said it is very possible that the ICS 9701C-03 or -11 14 pin could be replacements for CH9007D...

😮 kudos for fantastic googling skills. AV9107C-03 almost fits treeman diagram except for
- shorted ground and supply 😀 looks like error when drawing
- logic levels being flipped?

from diagram we have
FSO = 1
FS3 = 0
FS1/FS2 jumpered both to same logic levels. 1-2 1-2 giving 33MHz, 2-3 2-3 giving 25MHz

From the datasheet we would need
FSO = 0
FS3 = 1
FS1/FS2 1/0 25MHz and 0/1 33MHz

treeman remove both jumpers and buzz every pin once again, this time write down exactly which jumper block pins (JP4 1|2|3, JP5 1|2|3) are connected to which chip pins.

Jp5
pin 1 - > pin 3 4 5 9 10 12 14
pin 2 - > pin 2
pin 3 - > pin 3 4 5 9 10 12 14

Jp4
pin 1 - > pin 3 4 5 9 10 12 14
pin 2 - > pin 1
pin 3 - > pin 3 4 5 9 10 12 14

chrontel chip
pin 7 - > crystal leg1
pin 8 - > crystal leg2

pin 13 - > 14mhz

That is effing bizarre ! No way that is going to work. in either position of the jumpers it would be the same !
Typically middle pins is similar to yours but the outside pins: one must got to +V and other to Ground typically.....
added: check pin 12 of the Chrontel, see if 0 ohms to the +5v on the PSU connection.
check pin 5 and see if it is Ground at the PSU connection. see here if you need reference: https://fs3.kel.si/instructions/atpwr-pinout.htm

could this be a short inside the chip?
the logic of this setup seems illogical to me

treeman wrote on 2023-03-04, 00:32:

could this be a short inside the chip?
the logic of this setup seems illogical to me

Yes could be. Can you check the "added" stuff I put in above post ?