The frontside bus frequency of 33 MHz povided by the mainboard is doubled inside the CPU. So it effectively runs at 66 MHz. DX2 stands for "x2".

Thanks but at the cpu socket it reads 50 mhz not 66mhz. Im not seeing 66mhz anywhere. I actually take out the cpu to test it .

majestyk wrote on 2022-11-17, 07:41:

The frontside bus frequency of 33 MHz povided by the mainboard is doubled inside the CPU. So it effectively runs at 66 MHz. DX2 stands for "x2".

After a lot of fiddling I found that it is a bit different. In your case the socket receives the 50MHz from the crystal, this goes to CLK2. In a normal 'DX' processor this frequency is halved (to 25MHz), in the DX2 it is not halved (maintaining the 50MHz). The 33MHz crystal is probably for the bus, not the processor (as it would halve to 16MHz if led to the socket). If your mobo was meant to run at 33MHz, you would have a 66MHz crystal.

But the board has settings for 33 MHz and DX2 CPUS.
JP28 must be closed for DX2, in the picture it´s open (if it´s the one at the bottom right besides the empty CPU socket).

Where are the jumpers inside the blue box? Are they set for DX/DX2?

vacatedboat wrote on 2022-11-17, 07:56:

...I´m not seeing 66mhz anywhere. I actually take out the cpu to test it .

AFAIK 66 MHz are present inside the CPU only, you cannot measure the doubled frequecy at any pin of the socket.

Y2 is definitely the oscillator for the clock generator. By dividing and multiplying it can generate a multitude of frequencies with a given oscillator.

Many mainboards have a jumper for selecting if the generated clock frequency is passed to the CPU directly or if it is divided by 2. The CPU can then, if it´s a DX2, double it again.

Make sure there are no loose / corroded jumpers at the frequency jumper block.

Yes i have closed jp28 and also done jp7 and jp8 as per the table. That was my error the photo was taken on the weekend

What does the bios report at boot?

Blavius wrote on 2022-11-17, 07:57:

After a lot of fiddling I found that it is a bit different. In your case the socket receives the 50MHz from the crystal, this goes to CLK2. In a normal 'DX' processor this frequency is halved (to 25MHz), in the DX2 it is not halved (maintaining the 50MHz).

The standard 486 and 486DX pin doesn't have a CLK2 pin (which is named that way, because it receives two times the clock speed). A CLK2 pin is present on 286, 386 and some special early low-power 486 chips. The ordinary 486 processor has a CLK pin instead and expects the actual processor clock at that pin. So the 486DX-25 gets 25MHz, the 486DX33 get 33MHz. The DX2 version of processors contain an internal clock doubler, so they receive half of the clock. If there is in fact 50MHz at the socket of a 486DX2, the processor is operating at 100 MHz.

vacatedboat wrote on 2022-11-17, 07:38:

So the cystal near the isa slots at back. Barrel shaped read 33mhz.

You likely mis-read that. That crystal is for the real-time clock, and is supposed to read 32.768Hz, i.e. approximately 33kHz, not 33MHz.

vacatedboat wrote on 2022-11-17, 07:38:

The cystal in the middle of the MB read 50mhz and the socket clk read 50 mhz.

This is suprising. Your board uses a clock synthesizer chip, the small Macronix chip (MX logo) between the bigger UMC chips. JP9/JP10/JP24 configure that chip to a certain ratio between the crystal frequency and the processor frequency. The crystal frequency usually is 14.318 MHz, because that clock is required to be on the ISA slots for historical reasons. Please make sure you use proper grounding when trying to read clocks that high. If you don't properly ground the input of the handheld scope, you will find line frequency, i.e. 50Hz in European countries everywhere. As the jumpers are correctly set for 33MHz, I suspect that your 50MHz measurement is errorneous.

mkarcher wrote on 2022-11-17, 13:14:

The standard 486 and 486DX pin doesn't have a CLK2 pin (which is named that way, because it receives two times the clock speed). A CLK2 pin is present on 286, 386 and some special early low-power 486 chips. The ordinary 486 processor has a CLK pin instead and expects the actual processor clock at that pin. So the 486DX-25 gets 25MHz, the 486DX33 get 33MHz. The DX2 version of processors contain an internal clock doubler, so they receive half of the clock. If there is in fact 50MHz at the socket of a 486DX2, the processor is operating at 100 MHz.

Ah, ok I'm confused with the 386 then, thanks for the correction.

H3nrik V! wrote on 2022-11-17, 09:43:

What does the bios report at boot?

It reports correct 486 dx2. Just trying to understand diagnostic measures of frequency etc.

mkarcher wrote on 2022-11-17, 13:14:

Blavius wrote on 2022-11-17, 07:57:

After a lot of fiddling I found that it is a bit different. In your case the socket receives the 50MHz from the crystal, this goes to CLK2. In a normal 'DX' processor this frequency is halved (to 25MHz), in the DX2 it is not halved (maintaining the 50MHz).

The standard 486 and 486DX pin doesn't have a CLK2 pin (which is named that way, because it receives two times the clock speed). A CLK2 pin is present on 286, 386 and some special early low-power 486 chips. The ordinary 486 processor has a CLK pin instead and expects the actual processor clock at that pin. So the 486DX-25 gets 25MHz, the 486DX33 get 33MHz. The DX2 version of processors contain an internal clock doubler, so they receive half of the clock. If there is in fact 50MHz at the socket of a 486DX2, the processor is operating at 100 MHz.

That might explain why it was hot to touch after a few minutes. I have a heatsink for it though. Thanks
Should i be worried? Can i change it back to 33 mhz?

mkarcher wrote on 2022-11-17, 13:22:

You likely mis-read that. That crystal is for the real-time clock, and is supposed to read 32.768Hz, i.e. approximately 33kHz, n […]
Show full quote

vacatedboat wrote on 2022-11-17, 07:38:

So the cystal near the isa slots at back. Barrel shaped read 33mhz.

You likely mis-read that. That crystal is for the real-time clock, and is supposed to read 32.768Hz, i.e. approximately 33kHz, not 33MHz.

vacatedboat wrote on 2022-11-17, 07:38:

The cystal in the middle of the MB read 50mhz and the socket clk read 50 mhz.

This is suprising. Your board uses a clock synthesizer chip, the small Macronix chip (MX logo) between the bigger UMC chips. JP9/JP10/JP24 configure that chip to a certain ratio between the crystal frequency and the processor frequency. The crystal frequency usually is 14.318 MHz, because that clock is required to be on the ISA slots for historical reasons. Please make sure you use proper grounding when trying to read clocks that high. If you don't properly ground the input of the handheld scope, you will find line frequency, i.e. 50Hz in European countries everywhere. As the jumpers are correctly set for 33MHz, I suspect that your 50MHz measurement is errorneous.

Yes im in Oz a PAL region.
The gounding i try to use is the keyboard connector.

vacatedboat wrote on 2022-11-17, 19:38:

mkarcher wrote on 2022-11-17, 13:14:

Blavius wrote on 2022-11-17, 07:57:

After a lot of fiddling I found that it is a bit different. In your case the socket receives the 50MHz from the crystal, this goes to CLK2. In a normal 'DX' processor this frequency is halved (to 25MHz), in the DX2 it is not halved (maintaining the 50MHz).

The standard 486 and 486DX pin doesn't have a CLK2 pin (which is named that way, because it receives two times the clock speed). A CLK2 pin is present on 286, 386 and some special early low-power 486 chips. The ordinary 486 processor has a CLK pin instead and expects the actual processor clock at that pin. So the 486DX-25 gets 25MHz, the 486DX33 get 33MHz. The DX2 version of processors contain an internal clock doubler, so they receive half of the clock. If there is in fact 50MHz at the socket of a 486DX2, the processor is operating at 100 MHz.

That might explain why it was hot to touch after a few minutes. I have a heatsink for it though. Thanks
Should i be worried? Can i change it back to 33 mhz?

Until proven that the processor is really operating at 100MHz, I doubt it. Most 5V Intel DX2 processors barely overclock to 80MHz, even if heatsinked. Running a 486DX2 without heatsink at 66MHz can work with enough airflow. With a properly sized heatsink, convection cooling usually is enough, and with a heatsink/fan combination you are generally fine. I mistrust your 50MHz measurement, because you shouldn't see the same frequency at the crystal and at the processor, but the MX chip is supposed to convert 14.318MHz into either 33MHz or 50MHz (or some other frequencies). You have the correct jumpering for 33MHz on your board, so very likely your board correctly operates at 33MHz and your processor operates at 66MHz.

vacatedboat wrote on 2022-11-17, 19:41:

mkarcher wrote on 2022-11-17, 13:22:

You likely mis-read that. That crystal is for the real-time clock, and is supposed to read 32.768Hz, i.e. approximately 33kHz, n […]
Show full quote

vacatedboat wrote on 2022-11-17, 07:38:

So the cystal near the isa slots at back. Barrel shaped read 33mhz.

You likely mis-read that. That crystal is for the real-time clock, and is supposed to read 32.768Hz, i.e. approximately 33kHz, not 33MHz.

vacatedboat wrote on 2022-11-17, 07:38:

Yes im in Oz a PAL region.
The gounding i try to use is the keyboard connector.

This usually is a solid ground point - if it is connected to ground at all. What kind of handheld scope do you use? Do you know whether it can measure as high as 33MHz at all? Do you have a probe that you can switch between "x1" and "x10" mode? Usually, in x1 mode, you don't get good results above 5MHz, and if you directly probe the connections of a crystal, the probe loads the crystal so much that the dampening causes the oscillation to stop. You need to use x10 probing to probe oscillations at crystals and to measure signals faster than around 5 MHz.

vacatedboat wrote on 2022-11-17, 19:36:

H3nrik V! wrote on 2022-11-17, 09:43:

What does the bios report at boot?

It reports correct 486 dx2. Just trying to understand diagnostic measures of frequency etc.

What about frequency. Both 50 and 66 MHz Intel DX2s were available (and 40, but they're rare)

Thats at least re assuring. I might check the 3 jumpers at jp9, 10 and 24 to make sure they are making contact correctly and are all closed. Will use my meter.