VOGONS


First post, by Wizard83

User metadata
Rank Newbie
Rank
Newbie

Hi everyone, this is my first post and the first thread on this wonderful forum 😀
I am a fan of retro hardware, I often buy material from flea markets, online or through friends.
I have an old CHIPS vga card mod. GW-VGA with F82C451 chip, taken from a batch of material.
Some components have been removed from this card as you can see from the photo, the ram chips, the 4 quartz and a transistor in the upper part of the card.
I'd like to restore it if possible, at least to try it out and use it on my 286, if it works.
I need some help, if someone had the same card and could tell me about the missing components.
I found on wikimedia a photo where you can see the ram chips and it is already something, but I have not yet found info on the other components.
https://upload.wikimedia.org/wikipedia/common … a/ac/CT_VGA.jpg
thanks in advance to anyone who can give me a hand!

Attachments

Reply 1 of 17, by Predator99

User metadata
Rank Oldbie
Rank
Oldbie

Found a better picture at least, but no values visible.

I think for the OSCs you need the same values as for the 16-bit cards with the same chip...25, 28 and 40 MHz
http://www.vgamuseum.info/index.php/trade/ite … ologies-f82c451

ctvga.jpg
Filename
ctvga.jpg
File size
602.39 KiB
Views
480 views
File license
Fair use/fair dealing exception

Reply 2 of 17, by Wizard83

User metadata
Rank Newbie
Rank
Newbie

I had already talked to a friend and he told me the same thing.
I looked for photos of similar 16bit cards, but you can't see the transistor type 🙁
I still keep looking. thank you very much for the help

Reply 3 of 17, by Jo22

User metadata
Rank l33t++
Rank
l33t++
Wizard83 wrote on 2021-01-09, 16:24:

I had already talked to a friend and he told me the same thing.
I looked for photos of similar 16bit cards, but you can't see the transistor type 🙁
I still keep looking. thank you very much for the help

It's just an idea, but..
If it was a normal transistor (bi-polar), then a a generic silicon NPN type might do.

In Europe, the BC548 (or 546) was/is one of the most popular generic purpose NPN transistors (and BC558 as PNP).
It is/was used as an electronic switch, as AF amp, for little RF oscillators and so on.
The 548 and its sisters have the typical Collector-Base-Emitter pinout, as well.

In the US, the ancient 2N2222 had the same position.
However, it was/is more advanced and more powerful than the BC548.
Also, if memory serves, the plastic versions of the 2N2222 have a different pinout than the original metal can 2N2222s.

Anyway, these are just ideas, assuming that the missing transistor is a normal NPN type.
Because, PNP types are usually a thing of the 1950s to 1970s, when Germanium was the primary material still.
They were used because schematics were derived from older valve technology, which was PNP style essentially.

"Time, it seems, doesn't flow. For some it's fast, for some it's slow.
In what to one race is no time at all, another race can rise and fall..." - The Minstrel

//My video channel//

Reply 4 of 17, by Wizard83

User metadata
Rank Newbie
Rank
Newbie

I have other isa video cards that have a 2n3904, do you think it could be the missing transistor?
unfortunately the BCE pinout of the transistor is not indicated on the pcb
the only thing that you can clearly see is that the pin on the left of the transistor is connected to + 5V through the green inductance located on the top right.

Reply 5 of 17, by mkarcher

User metadata
Rank Member
Rank
Member
Wizard83 wrote on 2021-01-09, 14:43:

I need some help, if someone had the same card and could tell me about the missing components.

I can give you a hint about Q1: I checked three ISA graphics card with a classic 28-pin DIP RAMDAC. All of them have a three-terminal device near the RAMDAC, and it is labelled "Q1", although it is not a transistor, but a reference chip. It's a good thing that it is a reference chip, because you need a regulated current sink on that pin to get a good image. Two if the three cards a I checked use a TL431 reference chip. The connection is: Pin on one edge ("pin 3") to 5V (decoupled using an inductor), pin on the other edge ("pin 1") connected to pin 4 of the RAMDAC (Iref), a resistor between the pin 1 and pin 2, and another resistor between pin 2 and -12V (contact B7 on the ISA bus). The two resistors I mentioned will be R20 and R21 on your card.

If the resistors are as I described, a TL431 should do as replacement.

Reply 6 of 17, by Jo22

User metadata
Rank l33t++
Rank
l33t++
mkarcher wrote on 2021-01-09, 19:32:

I can give you a hint about Q1: I checked three ISA graphics card with a classic 28-pin DIP RAMDAC. All of them have a three-terminal device near the RAMDAC, and it is labelled "Q1", although it is not a transistor, but a reference chip. It's a good thing that it is a reference chip, because you need a regulated current sink on that pin to get a good image. Two if the three cards a I checked use a TL431 reference chip. The connection is: Pin on one edge ("pin 3") to 5V (decoupled using an inductor), pin on the other edge ("pin 1") connected to pin 4 of the RAMDAC (Iref), a resistor between the pin 1 and pin 2, and another resistor between pin 2 and -12V (contact B7 on the ISA bus). The two resistors I mentioned will be R20 and R21 on your card.

Interesting. From what I remember, "Q" also refers to oscillators (quarz crystals; complete crystal oscillators are named "X" or "XTAL") or mixers . Sometimes "Q" also refers to transistors (though in my country, "T" is more often used for this).

Or more precisely, here, "Q" refers to transistors that act as a switch (digital), while "T" is used if a transformation happens, like an amplification (analogue).
Though in practice, hobbysts here may simply use "T" anyway.

Edit: Seems there was a lot of confusion throughout the times.
In old schematics, transistors were als called "TR" or "V" (Germany), while transformers also had the "T" symbol. So yeah, it's really a mess. 😅

mkarcher wrote on 2021-01-09, 19:32:

If the resistors are as I described, a TL431 should do as replacement.

Oh well, so it might be a shunt device that acts similar to a z-diode? I didn't expect this. 😅
I expected ISA era devices would rather use big regulators in the size of mosfets or a 7805. I'm glad you got involved. 🙂

"Time, it seems, doesn't flow. For some it's fast, for some it's slow.
In what to one race is no time at all, another race can rise and fall..." - The Minstrel

//My video channel//

Reply 7 of 17, by mkarcher

User metadata
Rank Member
Rank
Member
Jo22 wrote on 2021-01-10, 07:04:
mkarcher wrote on 2021-01-09, 19:32:

I can give you a hint about Q1: I checked three ISA graphics card with a classic 28-pin DIP RAMDAC. All of them have a three-terminal device near the RAMDAC, and it is labelled "Q1", although it is not a transistor, but a reference chip. It's a good thing that it is a reference chip, because you need a regulated current sink on that pin to get a good image. Two if the three cards a I checked use a TL431 reference chip. The connection is: Pin on one edge ("pin 3") to 5V (decoupled using an inductor), pin on the other edge ("pin 1") connected to pin 4 of the RAMDAC (Iref), a resistor between the pin 1 and pin 2, and another resistor between pin 2 and -12V (contact B7 on the ISA bus). The two resistors I mentioned will be R20 and R21 on your card.

Interesting. From what I remember, "Q" also refers to oscillators (quarz crystals; complete crystal oscillators are named "X" or "XTAL") or mixers . Sometimes "Q" also refers to transistors (though in my country, "T" is more often used for this).

I haven't seen Q for "quartz" yet, at least I don't remember. I know X or Y both for crystals and crystal oscillator packages, though. Sometimes the oscillators also carry the generic "U" or "IC" marking for integrated circuits. I know Q mostly for transistors in english schematics, whereas german schematics tend to use T.

Jo22 wrote on 2021-01-10, 07:04:
Or more precisely, here, "Q" refers to transistors that act as a switch (digital), while "T" is used if a transformation happens […]
Show full quote

Or more precisely, here, "Q" refers to transistors that act as a switch (digital), while "T" is used if a transformation happens, like an amplification (analogue).
Though in practice, hobbysts here may simply use "T" anyway.

Edit: Seems there was a lot of confusion throughout the times.
In old schematics, transistors were als called "TR" or "V" (Germany), while transformers also had the "T" symbol. So yeah, it's really a mess. 😅

That's interesting. I don't remember any schematics that differentiated between switching transistors and amplification transistors in their symbols. And I would expect "TR" to be a transformer more likely than a transistor. The "V" is an anachronism, and is most likely rooted in "V"alve as a term for vacuum tubes that got replaced by semiconductor transistors.

Jo22 wrote on 2021-01-10, 07:04:
mkarcher wrote on 2021-01-09, 19:32:

If the resistors are as I described, a TL431 should do as replacement.

Oh well, so it might be a shunt device that acts similar to a z-diode? I didn't expect this. 😅
I expected ISA era devices would rather use big regulators in the size of mosfets or a 7805. I'm glad you got involved. 🙂

The VGA RAMDAC doesn't want a voltage refernce, it wants a current reference. If I understand the datasheet correctly, for a typical VGA design, you need to sink 8.8mA from pin 4 (Iref) to get correct RGB levels. The voltage at Iref is not clearly defined and might depend on temperature and be slightly different on different copies of the same chip. The datasheet for the KDA0476 shows a version with an adjustable negative voltage regulator used in current regulator configuration (they don't spell out which one to use, but I guess the LM337 variant in TO92 case would be fine), and the third ISA VGA card I looked at uses a LM334 current reference.

The TL431 (and all of its many copy cats like LM431 and so on) is likely a lot cheaper. The circuit works like this: There is a reference resistor between the reference input and the anode (remember, the Z diode has the anode at the negative end!). The reference input is also connected to the Iref pin. The TL431 is high impedance at the reference input, so the whole current from Iref passes through the reference resistor. The reference resistor is chosen such that the desired current causes a drop of 2.45 volts on the reference resistor. If the current from Iref is too high, the TL431 starts conducting more current, so the voltage drop on the resistor between anode and -12V increases, so the voltage at the anode increases, so voltage (and current) on the reference resistor goes down. If the current from Iref is too low, the TL431 doesn't conduct any current from cathode to anode, the voltage at the anode goes down and thus the current over the reference resistor increases - this is how the regulation loop is closed in this circuit.

I wonder whether the vertical banding you see even on analog monitors (so it most likely isn't a sampling artifact there, although you might get a moiree between the dot clock and the shadow mask) might be caused by instability of this Iref regulator due to degraded caps. The voltage at Iref might be dependant on the color, so the step at the start of a line (from black blanking to background color) can be an excitation source that synchronizes the instability of the regulator to the horizontal frequency.

Reply 8 of 17, by Wizard83

User metadata
Rank Newbie
Rank
Newbie
mkarcher wrote on 2021-01-09, 19:32:
Wizard83 wrote on 2021-01-09, 14:43:

I need some help, if someone had the same card and could tell me about the missing components.

I can give you a hint about Q1: I checked three ISA graphics card with a classic 28-pin DIP RAMDAC. All of them have a three-terminal device near the RAMDAC, and it is labelled "Q1", although it is not a transistor, but a reference chip. It's a good thing that it is a reference chip, because you need a regulated current sink on that pin to get a good image. Two if the three cards a I checked use a TL431 reference chip. The connection is: Pin on one edge ("pin 3") to 5V (decoupled using an inductor), pin on the other edge ("pin 1") connected to pin 4 of the RAMDAC (Iref), a resistor between the pin 1 and pin 2, and another resistor between pin 2 and -12V (contact B7 on the ISA bus). The two resistors I mentioned will be R20 and R21 on your card.

If the resistors are as I described, a TL431 should do as replacement.

not exactly...
pin 1 is grounded via R20 (150 ohm) and to pin 2 is connected to 1 via R21 which is 75 ohm.
Pin 2 is conneted to N4 of RAMDAC, not 1

Attachments

Reply 9 of 17, by mkarcher

User metadata
Rank Member
Rank
Member
Wizard83 wrote on 2021-01-11, 10:39:
mkarcher wrote on 2021-01-09, 19:32:
Wizard83 wrote on 2021-01-09, 14:43:

I need some help, if someone had the same card and could tell me about the missing components.

I can give you a hint about Q1: I checked three ISA graphics card with a classic 28-pin DIP RAMDAC. All of them have a three-terminal device near the RAMDAC, and it is labelled "Q1", although it is not a transistor, but a reference chip. It's a good thing that it is a reference chip, because you need a regulated current sink on that pin to get a good image. Two if the three cards a I checked use a TL431 reference chip. The connection is: Pin on one edge ("pin 3") to 5V (decoupled using an inductor), pin on the other edge ("pin 1") connected to pin 4 of the RAMDAC (Iref), a resistor between the pin 1 and pin 2, and another resistor between pin 2 and -12V (contact B7 on the ISA bus). The two resistors I mentioned will be R20 and R21 on your card.

If the resistors are as I described, a TL431 should do as replacement.

not exactly...
pin 1 is grounded via R20 of 150ohm and to pin 2 via R21 which is 75ohm.

OK, thanks for the schematics. The point of this circuit is to draw around 8.8mA from Iref. Assuming all current from Iref passes through R21 (and no significant part passes through the unknown 3-terminal device), the voltage drop on R21 should be 75 ohm multiplied by 8.8mA, which yields 0.66 volts. This is sufficiently close to the base-emitter voltage of an NPN transistor, so any general purpose transistor should do if it has base on pin 2, collector on pin 3 and emitter on pin 1. This is true for the 2N2222 and the 2N3904. The european BC238 or BC548 would also fit, but you need to turn it 180° to swap pins 1 and 3.

Reply 10 of 17, by Thermalwrong

User metadata
Rank Oldbie
Rank
Oldbie

I have the 16 bit version of this card in my 386 sx PC, I can possibly help with your 8-bit card. This actually came to me with another card with the same chipset which didn't work, but I can't remember where I put that card right now. Some of the RAM got borrowed from that card as this one originally had display corruption. So either of the two RAM types in the picture should work.

The transistor on mine is a 2N3904

The crystals are:
Y1 - 40.00
Y2 - 28.322

I didn't get the values of the cx capacitors so I can pull it out again and check those if you need those values too.

Attachments

Last edited by Thermalwrong on 2021-01-13, 17:10. Edited 1 time in total.

Reply 11 of 17, by mkarcher

User metadata
Rank Member
Rank
Member
Thermalwrong wrote on 2021-01-11, 20:31:
I have the 16 bit version of this card in my 386 sx PC, I can possibly help with your 8-bit card. This actually came to me with […]
Show full quote

I have the 16 bit version of this card in my 386 sx PC, I can possibly help with your 8-bit card. This actually came to me with another card with the same chipset which didn't work, but I can't remember where I put that card right now. Some of the RAM got borrowed from that card as this one originally had display corruption. So either of the two RAM types in the picture should work.

The transistor on mine is a 2N3904

The crystals are:
Y1 - 40.00
Y2 - 28.322
Y3 - ???
Y4 - 25.00 ??

I didn't get the values of the cx capacitors so I can pull it out again and check those if you need those values too.

If the card is supposed to output 100% VGA compatible timing, Y4 should be 25.175MHz, not 25.000MHz. It usually doesn't matter if the timing is off by 1%, so 25.000 might also do and is likely easier to obtain. Hercules pulled a similar stunt on their graphics-capable MDA replacement card (the Hercules graphics card) by using 16.000MHz instead of 16.257MHz.

Reply 12 of 17, by Wizard83

User metadata
Rank Newbie
Rank
Newbie
Thermalwrong wrote on 2021-01-11, 20:31:
I have the 16 bit version of this card in my 386 sx PC, I can possibly help with your 8-bit card. This actually came to me with […]
Show full quote

I have the 16 bit version of this card in my 386 sx PC, I can possibly help with your 8-bit card. This actually came to me with another card with the same chipset which didn't work, but I can't remember where I put that card right now. Some of the RAM got borrowed from that card as this one originally had display corruption. So either of the two RAM types in the picture should work.

The transistor on mine is a 2N3904

The crystals are:
Y1 - 40.00
Y2 - 28.322
Y3 - ???
Y4 - 25.00 ??

I didn't get the values of the cx capacitors so I can pull it out again and check those if you need those values too.

thank u! the transistor must be that, I'll mount it tonight! thanks everyone for the help!
for the oscillators.... my board has 4, from left to right Y3, Y1, Y2, Y4
I don't know how many were originally present, I just see that ALL the via are all very damaged as if they had used a chisel and not a soldering iron to remove them!
actually I think 3 capacitors are also missing, CX4, CX2 and CX6, but I'm not sure... for what the welds are, I suspect that only the CX2 was present ...i ordered the ram chips from china, i think they should arrive within a month.
i ordered the ram chips from china, i think they should arrive within a month. i took TMS4464-10NL

Reply 13 of 17, by Thermalwrong

User metadata
Rank Oldbie
Rank
Oldbie

As far as I'm aware, all of these cards should have 4 crystals. I got a better picture of them now and we can see that:
Y1 = 40.000MHz
Y2 = 28.322MHz
Y3 = 25.175MHz
Y4 = 36.000MHz

I wish you luck in getting your card working 😀

Attachments

  • IMG_07491.jpg
    Filename
    IMG_07491.jpg
    File size
    999.91 KiB
    Views
    213 views
    File comment
    Crystals and transistor
    File license
    CC-BY-4.0
  • IMG_074812.jpg
    Filename
    IMG_074812.jpg
    File size
    936.79 KiB
    Views
    213 views
    File comment
    CX capacitor values and positions
    File license
    CC-BY-4.0

Reply 14 of 17, by Wizard83

User metadata
Rank Newbie
Rank
Newbie

thanks for the crystals, I try to look for them!
for the CX capacitors, mine has the 1-3-5 present and they are all 50pF
2-4-6 are missing, but I don't know if they were factory missing.
on yours I see the CX1 and 5 are 50pf and 3 is 15pF, 4 and 6 are not present. there is another one missing with no component code on pcb, but going by exclusion I think it is the cx2
so there is a small difference.
it looks like these are part of a CRC cell, correct?

Reply 15 of 17, by mkarcher

User metadata
Rank Member
Rank
Member
Wizard83 wrote on 2021-01-13, 17:44:

thanks for the crystals, I try to look for them!
for the CX capacitors, mine has the 1-3-5 present and they are all 50pF

it looks like these are part of a CRC cell, correct?

I never heard the term "CRC cell", and google isn't helpful in connecting "CRC cell" with oscillators. The crystals usually run driven by an inverter (or other kind of negative gain amplifier), with the two ends of the crystal connected to "oscillator out" and "oscillator in". One or both sides of the crystal have a capacitor to ground to help impedance matching. These capacitors are often called "load capacitors", and they have a small but measurable effect on the resonant frequency. This is why professional crystals are specified for a given frequency at a given load capacitance. As long as you are not trying to generate the TV color carrier frequency, or you want to run a real-time clock off the line frequency of your video card, you don't need to worry about the exact values of the capacitors. As long as you get a stable picture, the card is equipped with capacitors well enough.

I don't know how they switch the clock frequency on your card: Either they have four oscillator circuits and select the output of one of the four oscillators, or they use an analog multiplexer to connect one of the four crystals to a single oscillator. In the end, it doesn't matter how they do it, as long as the oscillation is stable.

Reply 16 of 17, by Wizard83

User metadata
Rank Newbie
Rank
Newbie
mkarcher wrote on 2021-01-13, 19:45:
Wizard83 wrote on 2021-01-13, 17:44:

thanks for the crystals, I try to look for them!
for the CX capacitors, mine has the 1-3-5 present and they are all 50pF

it looks like these are part of a CRC cell, correct?

I never heard the term "CRC cell", and google isn't helpful in connecting "CRC cell" with oscillators. The crystals usually run driven by an inverter (or other kind of negative gain amplifier), with the two ends of the crystal connected to "oscillator out" and "oscillator in". One or both sides of the crystal have a capacitor to ground to help impedance matching. These capacitors are often called "load capacitors", and they have a small but measurable effect on the resonant frequency. This is why professional crystals are specified for a given frequency at a given load capacitance. As long as you are not trying to generate the TV color carrier frequency, or you want to run a real-time clock off the line frequency of your video card, you don't need to worry about the exact values of the capacitors. As long as you get a stable picture, the card is equipped with capacitors well enough.

I don't know how they switch the clock frequency on your card: Either they have four oscillator circuits and select the output of one of the four oscillators, or they use an analog multiplexer to connect one of the four crystals to a single oscillator. In the end, it doesn't matter how they do it, as long as the oscillation is stable.

thank you very much for the explanation 😀
sorry, the CRC cell is actually an RC circuit:
https://en.wikipedia.org/wiki/RC_circuit

Reply 17 of 17, by mkarcher

User metadata
Rank Member
Rank
Member
Wizard83 wrote on 2021-01-19, 12:39:
thank you very much for the explanation :-) sorry, the CRC cell is actually an RC circuit: https://en.wikipedia.org/wiki/RC_circ […]
Show full quote
mkarcher wrote on 2021-01-13, 19:45:
Wizard83 wrote on 2021-01-13, 17:44:

thanks for the crystals, I try to look for them!
for the CX capacitors, mine has the 1-3-5 present and they are all 50pF

it looks like these are part of a CRC cell, correct?

I never heard the term "CRC cell", and google isn't helpful in connecting "CRC cell" with oscillators. The crystals usually run driven by an inverter (or other kind of negative gain amplifier), with the two ends of the crystal connected to "oscillator out" and "oscillator in". One or both sides of the crystal have a capacitor to ground to help impedance matching. These capacitors are often called "load capacitors", and they have a small but measurable effect on the resonant frequency. This is why professional crystals are specified for a given frequency at a given load capacitance. As long as you are not trying to generate the TV color carrier frequency, or you want to run a real-time clock off the line frequency of your video card, you don't need to worry about the exact values of the capacitors. As long as you get a stable picture, the card is equipped with capacitors well enough.

I don't know how they switch the clock frequency on your card: Either they have four oscillator circuits and select the output of one of the four oscillators, or they use an analog multiplexer to connect one of the four crystals to a single oscillator. In the end, it doesn't matter how they do it, as long as the oscillation is stable.

thank you very much for the explanation 😀
sorry, the CRC cell is actually an RC circuit:
https://en.wikipedia.org/wiki/RC_circuit

It's not an RC circuit. RC oscillators are not accurate enough for dot clocks of graphics cards. Instead, it is a crystal-based Pirece oscillator.