VOGONS


Chicony CH-498B voltage regulator mod help.

Topic actions

Reply 20 of 90, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

So Deksor's board does pretty close to 3.15V - 3.6V in 0.15V steps, and also 4V (does that sound right?). For Hoping's board, it looks like the R1 158k should go nearest JP8, then pick out 4 resistors for the next 4 positions to give whatever voltage is wanted, chosen by putting a jumper on 1-2, 3-4, 5-6 or 7-8 (never more than one, that would put the resistors in parallel, lower the value of R2 and raise the output voltage). The order of the resistors on Deksor's board looks a bit random, but that could just be to make routeing easier and the order on the jumper makes sense.

Probably best to check the output voltage without a CPU fitted the first time...

[edit to add...] Just thought to say, not sure what others think, but I'd be nervous about putting a potentiometer/trimmer in a CPU voltage regulation circuit, except maybe without a CPU fitted as a way to find out what resistance is needed to get a particular output. Feels like too much chance for noise or dirty contact on the pot to cause voltage spikes.

That said, they appear in PSUs, so can't be all that bad.

Last edited by snufkin on 2021-10-19, 18:28. Edited 1 time in total.

Reply 21 of 90, by Hoping

User metadata
Rank Member
Rank
Member

So everything depends on the resistor's value and I need to follow the traces that go from the jumpers to the resistors if I understand it well and use a 94.5k resistor in one place to have the 3.3v, maybe ignoring the 4.02v resistor and using the odder ones may be the best.
And the resistor right above c23 besides the jumper j8, the one with the thicker trace would be R1 158k.
Will it work?
And it's so important to use precision resistors, will the voltage fluctuate a lot.

Reply 22 of 90, by weedeewee

User metadata
Rank Oldbie
Rank
Oldbie
snufkin wrote on 2021-10-19, 18:13:
The order of the resistors on Deksor's board looks a bit random, but that could just be to make routeing easier and the order on […]
Show full quote

The order of the resistors on Deksor's board looks a bit random, but that could just be to make routeing easier and the order on the jumper makes sense.

Probably best to check the output voltage without a CPU fitted the first time...

[edit to add...] Just thought to say, not sure what others think, but I'd be nervous about putting a potentiometer/trimmer in a CPU voltage regulation circuit, except maybe without a CPU fitted as a way to find out what resistance is needed to get a particular output. Feels like too much chance for noise or dirty contact on the pot to cause voltage spikes.

That said, they appear in PSUs, so can't be all that bad.

Seems like the resistors are pretty in line, with R1 158k being closest to the board edge.
agree on the not using a potentiometer aside from determining value when guessing. Since formula is known, no guessing needed.
PSU's it's more for fine tuning, not general selection.

Hoping wrote on 2021-10-19, 18:25:
So everything depends on the resistor's value and I need to follow the traces that go from the jumpers to the resistors if I und […]
Show full quote

So everything depends on the resistor's value and I need to follow the traces that go from the jumpers to the resistors if I understand it well and use a 94.5k resistor in one place to have the 3.3v, maybe ignoring the 4.02v resistor and using the odder ones may be the best.
And the resistor right above c23 besides the jumper j8, the one with the thicker trace would be R1 158k.
Will it work?
And it's so important to use precision resistors, will the voltage fluctuate a lot.

You'll need at least two resistors. With one in the right spot and the other spot will be determined by the jumper J8
from looking at deksor's board, R1 would be closest to the board edge, but since your board is different... R1 should be between the ADJ pin of the MIC 29302 and I guess one of the jumper pins

Reply 23 of 90, by Deksor

User metadata
Rank l33t
Rank
l33t

By the way, I own the full manual for this board, it should come soon. The one in the uh19 page is incomplete and in a really bad quality.

Trying to identify old hardware ? Visit Ultimate Retro - Project's thread The Ultimate Retro project - a stason.org/TH99 alternative

Reply 24 of 90, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie
Hoping wrote on 2021-10-19, 18:25:
So everything depends on the resistor's value and I need to follow the traces that go from the jumpers to the resistors if I und […]
Show full quote

So everything depends on the resistor's value and I need to follow the traces that go from the jumpers to the resistors if I understand it well and use a 94.5k resistor in one place to have the 3.3v, maybe ignoring the 4.02v resistor and using the odder ones may be the best.
And the resistor right above c23 besides the jumper j8, the one with the thicker trace would be R1 158k.
Will it work?
And it's so important to use precision resistors, will the voltage fluctuate a lot.

That sounds right to me.

Precision resistors aren't about voltage fluctuating, more what the final output voltage is. They have a couple of benefits that I can think of. Firstly for manufacturing where you can't test the value of every resistor. If they used +/-10% resistors then they might get unlucky and have the output voltage be over or under by over 20%.

Secondly they allow more options for the R1/R2 fraction. Common 10% resistors are 10, 15, 22, 33, 47 & 68, so trying to find the right ratio is a problem. So even if you have a big bag of common resistors to hand and you can measure the actual resistances (assuming you have a multimeter), you're going to struggle to find two with a ratio of e.g. 1.78 needed for 3.45V. Closest I can see would be a 9% high 150k with a 9% low 100k, and you'd have to be pretty lucky to find any that far away from their nominal resistance.

Just looked up a few things. So E12 series resistors are 10% and come in a greater range. Closest match for 3.45V would be 150k with 82k, which would come out as 3.51V. So you'd only need to find a 150k reading slightly high or an 82k reading slightly low to get the right ratio.

Short answer if that if you have a big bag of resistors you might be able to find just the right ratio, but it's probably easier to get the precision ones.

[edit to... you know the drill, one day I'll actually figure out what I want to say] Might be a obvious, but when doing the first tests, did you move the jumpers on that 5V 3V jumper block? I assume that disconnects the CPU from the board 5V and connects it to the regulator output instead.

Reply 25 of 90, by Hoping

User metadata
Rank Member
Rank
Member

Thanks a lot for the clear explanation about the precision resistors.😀
Back then, I left it because I really didn't know what I was doing and of course didn't want to damage the board. And what embarrasses me the most today is that I think I measured the voltage on the wrong pins of the socket, never tried to put the CPU without being sure about the voltage.
And yes, I'm sure I moved the 5v-3v jumpers block, but being me, anything could be possible.
Now I need to look for the components.
My other curiosity is if it will be better to use a MIC29152BT or a MIC29302 like I used, I have chosen it because of the 3A rating, but I guess that a 486 doesn't need that much and the only benefit would be less heat produced, or I'm wrong.

Reply 26 of 90, by weedeewee

User metadata
Rank Oldbie
Rank
Oldbie
Hoping wrote on 2021-10-19, 19:34:

My other curiosity is if it will be better to use a MIC29152BT or a MIC29302 like I used, I have chosen it because of the 3A rating, but I guess that a 486 doesn't need that much and the only benefit would be less heat produced, or I'm wrong.

you're wrong. 😀
anyway, the amount of heat produced by the regulator is dependent upon the voltage over the regulator multiplied by the current being drawn by the CPU.

The ratings of the regulators just are their maximum values before they will break.

the 3A ratings should allow you to run a cpu that consumes 3v3 * 3 ~ 10 Watt

Reply 27 of 90, by Hoping

User metadata
Rank Member
Rank
Member

So if the 486 dx4 100 consumes 5.22w maximum (cpu-world) and the mic 29152bt can give 3v3*1.5 around 4.95w, something is wrong. So I'll need the Mic 29302, right? What power rating for the resistors? Most ones I've seen are 400mw, is this low?
But I'm not used to this, and I'm having trouble to find the resistors with the correct values (:, but with patience and time maybe I'll get then.
Or maybe look for other values that are available, using the formula posted above (1.24*(1+R1/R2)). I guess that if this formula is correct the values that Chicony used doesn't need to be the only ones that works, R1 is fixed and R2 is the variable.
For example, 1.24*(1+160/95)=3.328, very close to 3.3.
Again if I'm right, never did this.

Reply 28 of 90, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie
Hoping wrote on 2021-10-19, 20:44:
So if the 486 dx4 100 consumes 5.22w maximum (cpu-world) and the mic 29152bt can give 3v3*1.5 around 4.95w, something is wrong. […]
Show full quote

So if the 486 dx4 100 consumes 5.22w maximum (cpu-world) and the mic 29152bt can give 3v3*1.5 around 4.95w, something is wrong. So I'll need the Mic 29302, right? What power rating for the resistors? Most ones I've seen are 400mw, is this low?
But I'm not used to this, and I'm having trouble to find the resistors with the correct values (:, but with patience and time maybe I'll get then.
Or maybe look for other values that are available, using the formula posted above (1.24*(1+R1/R2)). I guess that if this formula is correct the values that Chicony used doesn't need to be the only ones that works, R1 is fixed and R2 is the variable.
For example, 1.24*(1+160/95)=3.328, very close to 3.3.
Again if I'm right, never did this.

Let's see. What's the current through the CPU? Power=Current * Voltage (P=IV). Max power is 5.22W and Voltage is 3.3V, so current = 1.6A (on average). So the first thing is that the regulator needs to be able to supply that. The 29152 datasheet says max current of 1.5A. So that could be a problem, certainly borderline if the CPU is maxed out. The 29302 can supply 3A so that'd be ok. Except...

There's also the power lost in the regulator to be considered. The input voltage is 5V, the output 3.3V, so it's dropping 1.7V. P=IV=1.6A * 1.7V = 2.7W being burnt by the regulator.

The datasheet gives a formula for working out how good the heatsink needs to be: (TjMax - Tambient)/P - (Junction to Case + Case to Sink). Junction to Case for TO-220 is 2C/W (from datasheet), Case to Sink might be up to 2C/W (suggested by datasheet)

TjMax is 125C, and let's put Tambient at 50C (inside case, limited airflow). So (125-50)/2.7 - (2+2) = 28-4 = 24 C/W

So the heatsink temperature must only rise by 24C per Watt emitted (that's a measure of thermal resistance, the lower the better as it means power can be dumped without raising temperature much). 24C/W is probably lower than can be reached using just the copper pad on the PCB surface, so there's a chance of the regulator overheating if the CPU is drawing its full current for too long. It'd be nice to see a heatsink on the regulator to definitely get the thermal resistance below 24C/W.

The resistors aren't actually handling much current, I think mA, so their power handling isn't a problem. They're just providing a feedback signal to the regulator.

Assuming I've gotten things right (I think the voltages based on Deskor's board look about right, it'll be good to see what the manual says), then you're right that the value don't need to be the same. I'd probably stay around the same values though, so keeping R1 between 100k and 200k. That said, the closest I can find to 3.3V using common resistor values is R1=56, R2=33, giving 3.34. So maybe 56k and 33k?

Reply 29 of 90, by Robin4

User metadata
Rank l33t
Rank
l33t
weedeewee wrote on 2021-10-19, 16:57:
R20 ? brown red blue black violet ? 126 R21 brown black red orange brown ? 102K R.. brown red brown silver silver that don's m […]
Show full quote

R20 ? brown red blue black violet ? 126
R21 brown black red orange brown ? 102K
R.. brown red brown silver silver that don's make sense other way round and silver is gray sigh... 881

Wait a minute... there's only place for 5 on OP's board, but deksor's board has six.

silver red green red brown ... gray again 82.5k
brown red green ? white ... can't tell what color that missing one is... looks like lime but ... does not compute 😉 I'm guessing it's gray. and reversed 98.5K
brown green silver orange brown 158k also gray not silver :-p

edit :
violet black blue red brown 70.6K
brown black red orange brown 102k
gray gray brown red brown 88.1K
gray red green red brown 82.5K
white gray yellow green red brown 94.5K
brown green gray orange brown 158K
edit2:
reversed
126
132
12100M <- does not make sense
12K5
12500M <- also does not make sense
13M8 <- also does not make sense

edit : lime = yellow not gray
deksor's board has 5 voltage jumper settings, Op's board only has 4 :-p

Where do you want buy those resistors.. The value`s are hard to find now days.. Or need to find old electronics and reuse these on the motherboard in question.

~ At least it can do black and white~

Reply 30 of 90, by Hoping

User metadata
Rank Member
Rank
Member
snufkin wrote on 2021-10-19, 21:42:
Let's see. What's the current through the CPU? Power=Current * Voltage (P=IV). Max power is 5.22W and Voltage is 3.3V, so curr […]
Show full quote

Let's see. What's the current through the CPU? Power=Current * Voltage (P=IV). Max power is 5.22W and Voltage is 3.3V, so current = 1.6A (on average). So the first thing is that the regulator needs to be able to supply that. The 29152 datasheet says max current of 1.5A. So that could be a problem, certainly borderline if the CPU is maxed out. The 29302 can supply 3A so that'd be ok. Except...

There's also the power lost in the regulator to be considered. The input voltage is 5V, the output 3.3V, so it's dropping 1.7V. P=IV=1.6A * 1.7V = 2.7W being burnt by the regulator.

The datasheet gives a formula for working out how good the heatsink needs to be: (TjMax - Tambient)/P - (Junction to Case + Case to Sink). Junction to Case for TO-220 is 2C/W (from datasheet), Case to Sink might be up to 2C/W (suggested by datasheet)

TjMax is 125C, and let's put Tambient at 50C (inside case, limited airflow). So (125-50)/2.7 - (2+2) = 28-4 = 24 C/W

So the heatsink temperature must only rise by 24C per Watt emitted (that's a measure of thermal resistance, the lower the better as it means power can be dumped without raising temperature much). 24C/W is probably lower than can be reached using just the copper pad on the PCB surface, so there's a chance of the regulator overheating if the CPU is drawing its full current for too long. It'd be nice to see a heatsink on the regulator to definitely get the thermal resistance below 24C/W.

The resistors aren't actually handling much current, I think mA, so their power handling isn't a problem. They're just providing a feedback signal to the regulator.

Assuming I've gotten things right (I think the voltages based on Deskor's board look about right, it'll be good to see what the manual says), then you're right that the value don't need to be the same. I'd probably stay around the same values though, so keeping R1 between 100k and 200k. That said, the closest I can find to 3.3V using common resistor values is R1=56, R2=33, giving 3.34. So maybe 56k and 33k?

I don't know what to say, I'm impressed, and the explanation is very good and clear.
Thanks a lot.
A more clear manual can be found here https://www.elhvb.com/mboards/chicony/ch498b/CH-498B.html.
I guess it was the base to the PDF on the link provided be Deskor.

Robin4 wrote on 2021-10-19, 21:58:

Where do you want buy those resistors.. The value`s are hard to find now days.. Or need to find old electronics and reuse these on the motherboard in question.

Like I said I have very little experience with this, I only bought components on the physical store and the stores on my town are useless so I have to look for then online but the big ones, rs,mouser,farnell they have very high shipping costs here( Spain).

Reply 31 of 90, by Deksor

User metadata
Rank l33t
Rank
l33t

This manual is the same I used to create the pdf.
However like I said I own the entire manual, I scanned it, I now need to assemble it into a pdf (and perhaps ocr it)

Edit : it's online on the win3x page !

Trying to identify old hardware ? Visit Ultimate Retro - Project's thread The Ultimate Retro project - a stason.org/TH99 alternative

Reply 32 of 90, by Robin4

User metadata
Rank l33t
Rank
l33t
Deksor wrote on 2021-10-20, 15:37:

This manual is the same I used to create the pdf.
However like I said I own the entire manual, I scanned it, I now need to assemble it into a pdf (and perhaps ocr it)

Edit : it's online on the win3x page !

Do you also have that 3.07 version bios? I see you made a pointer to it. But it doesnt seems to be uploaded to your site.

~ At least it can do black and white~

Reply 33 of 90, by Deksor

User metadata
Rank l33t
Rank
l33t

No, somebody probably uploaded a picture or the post string but never posted that version online, so it will be missing until someone finds it and sends it to us

Trying to identify old hardware ? Visit Ultimate Retro - Project's thread The Ultimate Retro project - a stason.org/TH99 alternative

Reply 34 of 90, by Hoping

User metadata
Rank Member
Rank
Member

I am going to have to buy all the components in China since the shipping costs of the large online stores are an insult to intelligence, and they do not have the MIC29302WT that I need in stock.
For this reason it will take a long time to start the tests. I guess about a month, which is usually around here for shipments from China.

Reply 35 of 90, by Hoping

User metadata
Rank Member
Rank
Member

Does anyone have experience on undervolting an 486?
Because if all goes well I will have four voltage options that I'll can choose according to the resistors I use. Personally, whenever I use a component at its standard frequency, I like to look for the minimum voltage at which it is stable.
I would appreciate knowing from experiences in this regard, what voltage would be the minimum with which a 486 of 3.3 volts boots.

Reply 36 of 90, by brunocrod

User metadata
Rank Newbie
Rank
Newbie

Don't get it. I have the same mb, but it is 1.0. And there is a difference. So strange. Is yours really an original 1.0 or a modified one? My mb has the capacitor C71 installed and doesn't have this white bracket from the socket installed. I have owned mine for 27 years, never mod, I am the only owner. I Never knew there was another type of it with the same version.

Attachments

Reply 37 of 90, by Hoping

User metadata
Rank Member
Rank
Member

The only mod present on my photos is the MIC voltage regulator, I did not remove the C71 capacitor.
My board has a ZIF cpu socket maybe your board has a LIF socket. When I continue with this proyect I wil try to do better photos.
Edit: Maybe you can check your board's bios version to see if it's diferent fron the one my board has.

Reply 38 of 90, by Deksor

User metadata
Rank l33t
Rank
l33t

My board is ver 2.0 and I didn't make any changes to it either (I'm definitely not the original owner though).

I believe your PCB and his PCB are rev 1.0, but some of the components may have changed.
Looking at the markings around your LIF socket and the fact we have a proper socket 3 makes me think that they might have had issues with availability of socket 3 ZIF so they simply went back to LIF socket instead. There are many boards with this kind of difference, and yet still the room to install a ZIF socket.

Trying to identify old hardware ? Visit Ultimate Retro - Project's thread The Ultimate Retro project - a stason.org/TH99 alternative

Reply 39 of 90, by brunocrod

User metadata
Rank Newbie
Rank
Newbie
Hoping wrote on 2021-11-06, 13:58:

The only mod present on my photos is the MIC voltage regulator, I did not remove the C71 capacitor.
My board has a ZIF cpu socket maybe your board has a LIF socket. When I continue with this proyect I wil try to do better photos.
Edit: Maybe you can check your board's bios version to see if it's diferent fron the one my board has.

I will check later