VOGONS


Reply 20 of 59, by canthearu

User metadata
Rank Oldbie
Rank
Oldbie
filurkatten wrote on 2021-12-30, 23:34:

I have recapped the abit board and it starts up but shows both my 450MHz cpus as 300MHz. Guessing the fsb is set to 66?

Also, CT4810 or CT4670 for windows pci audio?

Yep, Need to set FSB to 100mhz to get 450mhz output from PII-450 CPUs.

CT4670 (Sound Blaster Live!) is generally superior for features compared to the CT4810 (Sound Blaster PCI). Audio quality, might actually be slightly better on the CT4810, but neither is particularly bad.

DOS FM Emulation is better on the Live is better (although still not great) , and you get Environmental Audio under windows (EAX)

Reply 21 of 59, by filurkatten

User metadata
Rank Member
Rank
Member

Got the bios updated and now I can select higher cpu multiplier. From x4.5 to X9.
Next step is to install windows and test the gpus, soundcard and a ethernet card.
Got around 6-8 IDE harddrives so hopefully one is working. 😀

Reply 22 of 59, by filurkatten

User metadata
Rank Member
Rank
Member

Still waiting to meet up with my friend to test the new CPU.
Got a HDD working, but i am looking into using a more modern alternative like a IDE to SATA adapter and a SSD.
Now I got another question regarding RAM.
So, if I get the P3 850 working I would like to oc it a little bit to be closer to 1GHz and from what I can read good RAM is of essence.
I dont have that much to choose from at the moment but I'm willing to buy if needed.
Seems like 3x 128M or 1x 256M + 1x 128M is the way to go and I correct with this assumption? Or should I go for 512M with WinME?
Then CL2 seems to be the better option, but I dont think I got any RAM that is CL2 (is it possible to OC the ram to use CL2 instead?)
And if I OC the CPU then PC133 is the better option and as fast as possible? (7 or under?)

So, this is the RAM I got available, I have written label if any and name and part´number on chip and how many chips on the stick.
I think I got the values correct where I could find them.

|------------
Hyundai 64M PC133 CL3
8x GM72V66841ET75

|------------
StarRam 128M - These chips look different then the rest, they are small and square.
8x 16MX8F2 0147

|------------
Micro Memory 128M PC133 CL3
LHT - Bigger chips that lays on the side
4x 48LC16M16LF -75E

|------------
XELO 128M PC133
8x TM3V8M1618U-7A

|------------
Toshiba 128M PC100
16x TC59S6408BFT-80

|------------
TI 64M PC100
8x TMS664814DGE -8A

|------------
2x EliteMT 128M PC133
8x M12L64164A -8T

Is this RAM any good or should I look for some better quality brands and values?
From just looking at the RAM the XELO seems good but I only have 1. The EliteMT I have 2 but it's a bit slower.
Thanks in advance!

Reply 23 of 59, by Meatball

User metadata
Rank Oldbie
Rank
Oldbie

Go with CL2 RAM if you feel like sparing a few dimes. You should not find any price difference between CL2 and CL3 in the PC66/100/133 market. You don't have to use it at CL2, but the guaranteed performance is there if you need it and/or for overclocking cushions. Many sellers won't mention (or know to mention) CL levels, so check everything over manually on the RAM stick before purchasing.

I primarily go with Micron/Crucial, but Nanya, Samsung, & Hynix/Hyundai are all great and I own all of those brands. With that said, I haven't had any trouble with off-brand RAM. You should be OK with whatever is available. Personal preference, mostly, after verification of capabilities you are shopping.

Last edited by Snover on 2024-02-15, 21:33. Edited 2 times in total.
Reason: Restoring revision 136982

Reply 24 of 59, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

Pretty sure (it's worked for me) that anything rated as PC133/CL3 will work as PC100/CL2 Datasheets for the actual DRAM chips usually have them rated for both. Plus if you start OCing the bus then if you start to get problems with the RAM then you know can always back off to CL3. If you really want speed then there are some PC133/CL2 as well. I think that which of PC133/CL3 (running at 133MHz Bus) or PC100/CL2 (running at 100MHz Bus) is faster will depend on what program is being run (CL2 has slightly better latency even at 100MHz, but PC133 has better data rate).

For testing I usually let memtest86 run all tests at least overnight, ideally 24 hours, with no errors. Had some RAM once that was fine overnight, but then started throwing occasional errors during the day which turned out to be temperature related.

Skimming a few datasheets, part numbers with 7A will be PC133/CL2, 75 will be PC100/CL2 or PC133/CL3, 8 will be PC100/CL3 (should be ok up to 125MHz@CL3). Can't find a datasheet for the exact part, but that Xelo stick might be PC133/CL2.

Reply 25 of 59, by filurkatten

User metadata
Rank Member
Rank
Member

Thanks for the info!
I tried out a couple of my RAM sticks but some where faulty so I think I'll try to find a 256 PC133/CL2 stick and then use my Xelo stick that was working.

Also found Win ME installed on one HDD so I installed nvidia drivers and voodoo 2 drivers to test that the cards where working correctly and I had no problem with any of them. 😀
Got SLI working with the Voodoo 2 aswell so I'm really happy.

Will meet up with my friend and get the 850MHz CPU later this week and hopefully it works with the Abit board.

So next up is getting my hands on new RAM and get a storage solution.

What do you guys recomend for storage? Ive read many uses CF cards but the bigger sizes are a bit to expensive for this project.
So I was thinking 2x ide-to-sata and 120GB SSD.
One drive partitioned with windows partition and one game partition.
The other drive with game CD Images.

Another question, how bad is the heat generated by the GPUs?
Should I mount a fan blowing over the Voodoo 2's?
Got a 3D printer so I could make some mounting brackets and mount it behind the GPU's for on the side.

Thanks again every one for the answers, I'm having a really good time building my computer, it takes some time with work and having kids at home but I'm getting there. 😀

Reply 26 of 59, by filurkatten

User metadata
Rank Member
Rank
Member

Switched cases to one with a 120mm intake fan in front and one 120mm exhaust fan in the back.
Hoping this will give good airflow.

EDIT
After hooking up the fans I noticed one wasnt spinning, the one connectod to the second fan header.
After measuring the pins I saw that the voltage only was 6V instead of 12V that it should be.

Do you guys have any suggestions how to troubleshoot this problem?
This pin header is reported as power fan in BIOS.
Also I get no rpm readings at all from my casefans in BIOS.
CPU fans reports thou.

Installed SpeedFan to check from within Windows, Temp 1 and 3 seems to give some reading, I think Temp 3 is the CPU thermistor. It was around 25-30 degree celcius.
But Temp 2 was at 127 degree celcius, guessing a sensor is broke or unused.
I think my board uses the winbond 83782D chip for the readings.
Looking at the manual it can only read two temps: "Hardware monitoring - Included fan speed, voltages, CPU and system
environment temperature"
So I guess Temp2 is unused.
Still one fan header with half the voltage and so fan speeds sensed gets me a little worried.

Reply 27 of 59, by filurkatten

User metadata
Rank Member
Rank
Member

Did some more experimenting with the fans, also tested another PSU (the one I'm using mainly is a modern PSU from around 2014 and the one I tested with was another beefy PSU from ~98).
The cpu fan header and one of the chassi fan headers return 12V correctly.
One chassi fan header (power fan in bios) is only at 6V.
None of my chassi fans returns any RPM readings on the working headers but the cpu 80mm fan does. So my guess is that the rpm is so low that the board doesnt pick up on it.

Around the troublesome 6V fan header I can find that 12V is present with my multimetet.
It's to hard thou to follow where the fan leads to check where the voltage is coming from, to many traces and the origin is obstructed by the header itself.

Cpu fan speed isnt changeable in BIOS.
I'm thinking that maybe I should downgrade the BIOS to see that it isnt just a bug.

Thou 6V instead of 12V is a bit concerning.
There is no caps atround this area that I have changed so I dont think thats the problem.

Anyone got any suggestions?
Is it common that fan headers break on these boards?
Only thing I can think is to desolder the header then follow the trace but I rather not if possible.

EDIT

I probed around some more and I think I might have found where the problem comes from.
This is the fan header that has the issue:
20220114-124104.jpg

The component Q7, the thick trace that goes to it's lower right leg is a 12V line, and that checks out and has the correct voltage level.
Then it's left lower leg and top tab is connected and goes to the fan header pin 2 with 6V.
It also connects down the line to R196, R195.
Some where in this circuit after the 12V has passed Q7 the voltage drops.

Reply 28 of 59, by filurkatten

User metadata
Rank Member
Rank
Member

Its the Q7 transistor.
I compared it to the other transistors by the other fan headers.
The Q7 one is shorted on the middle and bottom leg.
Guess I should remove it and or replace it if I can find a suitable replacement.
Its marking is P06 so its a amplifier if Im correct.

Reply 29 of 59, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

Possibly this: https://www.mouser.co.uk/datasheet/2/427/70229-1765204.pdf (Marking Code for TP0101T: POwll, w = Week Code, l = Lot Traceability)

I guess the body diode has gone short. P channel, so when the Gate (pin 1) is below Source (pin 2) then current can flow from Source to Drain (pin 3). Pretty much any PFET should work, maybe something like https://www.mouser.co.uk/datasheet/2/427/si23 … 7ds-1764782.pdf

[looks like if Q14 is off then R196 would pull the Gate of Q7 to 12V, so turn it off. If Q14 is on then it would pull one side of R195 to Ground. That would pull the other side of R195 and R196 to half way between 12V and 0V, so that'll be why you saw 6V, and I'd guess any attempt to actually draw any current would see that drop to 0V. So Q14 can be used to turn the fan off, I wonder if that allows for speed control]

Reply 30 of 59, by filurkatten

User metadata
Rank Member
Rank
Member

The BIOS doesnt have a fan control but it has the option to turn them off when in power saving mode so a switch makes sense.

When searching for P06 smd transistor I found that this part: PDTC124E had the marking P06.
Its an NPN resistor equipped transistor.

Also part SO2894 seems to had the marking P06.

What tells you that the part is a PFET?

I think its a zero in the marking, P zero six.
It also has 94 at a angle.

Reply 31 of 59, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

Ah, it was largely a guess based on what I found for P0 or PO. You measured the lower right pin (pin 2) as 12V, and that's a thick trace so is likely the power input. Pin 1 has a thin trace so will be the control signal, which leaves pin 3 as the output. That then probably goes to the +ve pin on the fan header (middle pin), through the fan motor, down to Ground (fan header pin 1). So when that device is conducting both the input (pin 2, 12V supply) and the output (pin 3 to fan header) will be at around 12V. When it's not conducting then pin 3 will go to 0V.

An NPN transistor (like the PDTC124E) will conduct when the base (pin 1) voltage is above the emitter (pin 2), which would mean the base would have to be more than 12V, which seems very unlikely. Plus, you measured pin 2 as 12V, so current would have to flow from the emitter to the collector, which is wrong for an NPN. So it looks like we're switch on the high side of the fan, not the low side. So it's either a PNP BJT or P-ch FET.

The SO2894 is a PNP transistor, so that seems right for where this transistor is in the circuit. Current flowing from the Emitter (pin 2) to the Base (pin 1) allows current to flow from Emitter to Collector (pin 3). But it has a maximum Vce of 12V, which is exactly what the voltage be if the fan is off, and a maximum current of 200mA which doesn't seem high enough, and it's unlikely for the designer to have chosen a part to run at its absolute maximum.

That TP0101T has a max Vds of 20V and Id of ~500mA, which seems more likely. Plus, by the time this board was designed (98/99?) FETs would have been common for switching.

So, my guess is a P-ch FET, with the marking being type PO, week code 6 (so week 46-50 according to Vishay's week code table), lot 94.

Could be wrong though. You can check (with board off) that pin 3 does connect to pin 2 of the header. That'll confirm if Q7 is P and not N type. Also if you remove Q7 you can measure (with the board on) the voltage at pin 1 when the fan is supposed to be on, and when it's in power saving mode and the fan off. Also, measure the pin 1 voltages on the other fan control transistors. A difference between BJT and FET is that, for p type, the Emitter-Base voltage of a BJT won't be very large (just a diode drop, so <1V), so pin 1 will be around 11V. But a FET can have a large Source-Gate voltage (in this case probably 6V) so will be around 6V.

Reply 32 of 59, by filurkatten

User metadata
Rank Member
Rank
Member

Great answer! Thanks you, much appriciated.

I desoldered the Q7 transistor and it was indeed shorted by pin 1 and 3.
Pin 3 on the board goes to middle fanheader pin.
When the board is powered without the transistor pin 1 gets 6.5V, pin 2 12V and pin 3 is disconnected so your assumptions where correct.

I'm a bit limited with what I can order from where I live if I order small quantities and if I want the component quick.
But I can get a hold of the component FDN358P that seems that it maybe could work.
What do you think?
Also can get FDN352AP and NDS352AP but FDN358P has a bit higher rated and has lower rdson.

I did also measure pin 1 on another fan header but ut was reading ~11v. But I guess its because its still in circuit.

Reply 33 of 59, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie
filurkatten wrote on 2022-01-15, 23:35:

I did also measure pin 1 on another fan header but ut was reading ~11v. But I guess its because its still in circuit.

Ah, that might mean I'm wrong and it was a PNP transistor, as that means the voltage from pin 2 Emitter (at 12V) to pin 1 Base (you've measured 11V) is only 1V, which looks like a Vbe diode drop. So you may have been correct with that PNP that you found, even if that does mean it's running at its limits.

Still, I think a P-ch should work, it's just switching a load, so it's not doing any complicated analogue stuff, and the pin 1 input is actively driven, so it's not going to leave the gate input floating. Maybe there'll be a problem with back-emf if turning the fan off, but a PNP would have a problem with that as well.

That FDN358P looks ok I think. Max Vgs and Vds are both above 12V, current is 1.5A. 0.2 ohm max Rds looks a little high, but P=I^2*R, so even if the fan draws 0.5A then the power will only be 0.5^2 * 0.2 = 50mW, so that's fine. Vth of max 3V should also be ok since it'll actually be around 6V. There's a resistor between Q7 and Q14, so a problem with Q7 shouldn't cause any damage to the rest of the board.

Reply 35 of 59, by shamino

User metadata
Rank l33t
Rank
l33t

I think having marginal transistors on those fan headers might have been common back then.
I have an Epox slot-1 board which had a similar problem. Assuming I had found the correct datasheet then it looked like the max current rating on it's originals was something like 0.2A (I don't remember the number exactly).

One of the headers was dead when I got the board, and the other died the first time I powered it up - because one of the fans was seized and I plugged it into a different header than the previous owner did.
The replacement was rated for way more current, something like 2A. I don't know how hard the circuit traces on a board of this era can actually be pushed before they'd burn though.

Reply 36 of 59, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie
filurkatten wrote on 2022-01-16, 02:22:

Should I be concerned that it was 6.5V at pin 1 after I removed the transistor when the other fan header was 11V?

Don't think so. I'm going to go through this mostly to convince myself that it sounds about right, I'll go on a bit...

So without the Q7 transistor fitted then the trace going to pin 1 is pulled through two resistors: R196 and R195 (both look to be 1k resistors, marked as 102, so 10 and 2 zeroes). R196 tries to pull it up to 12V, R195 tries to pull it it to whatever the voltage on pin 3 of Q14 is. I'm assuming that Q14 pin 3 is at 0V when the fan is on.

So with one side at 12V and the other at 0V, with equal value resistors, and no current flowing through pin 1 (because it's not connected to anything) the voltage at Q7 pin 1 ends up half way between the two, around about 6V (there'll be small differences depending on the exact value of the resistors and the 12V supply). If a FET is fitted there then still no current flows, since no DC current flows in or out of the gate (there is gate capacitance, so some current does move when it switches on or off), so the voltage will stay the same.

But it's different with a PNP transistor fitted. There's effectively a diode going from the Emitter (pin 2) to the Base (pin 1). As you increase the voltage across a diode then only a little current will flow until it reaches a threshold, at which point if you increase the voltage a little more then there will be a huge increase in the current. The Emitter is at 12V, so current can flow through the diode from pin 2 to pin1, and then it can flow through R195, in to Q14 and down to Ground. That current flowing through R195 will increase the voltage across it.

Mostly ignoring R196, if the Base is at 6V and the Emitter is a 12 V then a huge current would flow. Except that current would have to flow through R195, so the voltage across R195 increases, so the Base voltage increases. That reduces the Emitter-Base current. Very quickly an equilibrium will be reached between the current and voltage through/across the diode and the current and voltage through/across R195. Which will turn out to be that the voltage across the Emitter-Base diode will be around the turn on voltage for the diode, so around 1V. So you end up with the Emitter at 12V and the Base at 11V, with about 11mA (11V / 1k) flowing through R195. 1mA from R196 and 10mA through the diode. That 10mA allows a much larger current to flow from the Emitter to the Collector, and then to the fan.

So, with a FET you'd expect the Gate to be at around 6V because no current flows through the gate, so the voltage is set purely by the voltage divider of R195 and R196. With a BJT you'd expect the Base voltage to be around 11V because current is flowing out of the Base and then through R195, raising the voltage of the Base until it's a diode voltage drop below the Emitter.

From what Shamino says then it sounds like it was common to have the transistors controlling the fans be a bit marginal. I'm slightly worried that I'm missing something that means a FET won't work. But see how it goes.

Reply 37 of 59, by filurkatten

User metadata
Rank Member
Rank
Member

Shamino; good to hear that its a common issue so its not something else on the board that might have caused it. 😀

Snufkin; great explanaition.
I think I understand how it should work.
From what you wrote a FET should work I think and I put in an order for some.
Hopefully I can test it later this week.
Also looked for some PNP transistors as backup, I found one mmbt3906lt1g thats the closest one I can order without having to wait for months on the delivery or pay way more for shipping than the component itself.
It has similar values but is rated for 40V. Would that matter in this case?

I'll keep you updated with how the repair goes.

Reply 38 of 59, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie
filurkatten wrote on 2022-01-16, 17:00:

Also looked for some PNP transistors as backup, I found one mmbt3906lt1g thats the closest one I can order without having to wait for months on the delivery or pay way more for shipping than the component itself.
It has similar values but is rated for 40V. Would that matter in this case?

Still a maximum of 200mA, which seems low, but is probably the same as before. DC current gain drops pretty low for high currents out of the Collector (about 30 if drawing 100mA). But the Emitter-Base current should be around 10mA, so should still be ok. Good luck with the swap out, hope I've got this right.

Reply 39 of 59, by filurkatten

User metadata
Rank Member
Rank
Member

Havnt got the transistors yet but I did have time to test the 850MHz cpu.
After changing the multiplyer to X8.5 100fsb it worked and bios reported Pentium III with 850MHz 1.65V and windows started up. 😀

EDIT
So, the computer started to crash with the new cpu.
The temps where at around 40 degree celcius.
The cooler on the new cpu is a passive cooler with large blades and the old one had smaller blades but with a fan connected.
So first I reapplied some thermalpaste. Artic silver I think it was. I noticed that the cooler didnt have a conductive cooper plate or anything and it didnt have any standoffs between the cpu board and the cooler itself.
I proceeded and installed it again it it continued to crash at 40 degrees.
Then I thought that it must be that it didnt conduct properly from the cpu to the cooler so I tried the old cooler with the fan. Cooler Master brand. It has a copper plate between the cpu and the heatsink also standoffs so it lies flat.
Now everything is stable and after having Unreal on for a little while the cooler was at around 32 degree celcius.
I think it was a success.

I maybe will change the cpu fan to a more quiet one in the future.