Thanks for the summary, you put things in a rather more succinct readable form than I! 😉
Its a shame you cant quite fit more 18650 s in, I was rather hoping an off the shelf 2s2p pack would go in! I mentioned the charge controller supports 3 cells, with a circuit mod, but a 2s2p pack (with inbuilt protection & ideally charge leveling), if the case can be modded, should work OK with the existing charge controller etc.

I was looking at the GD542* datasheets last night trying to work out if it was easy to do a swap like you mentioned, you've saved me some work there, thanks! Did you have to do a specific ROM for the GD5428 with 512kKB?
Is the GD5426 also a drop in replacement?

The coin cell will help of course, like when the main battery is flat... however on the pocket386 the lack of one was not the main issue causing low of date/time /BIOS resets.- I know, I tried putting one on ... and Ive had a 'scope hooked up to that supply to the M6117, it was rock solid when powering on/off/plugging in external supplies. Others on the net also report a battery did not help.
This is why i tried a proper PSU monitor IC to generate the PWG signal. I'm not sure why this isnt working 100% on the BOOK386, I can only assume its the extra delay i put in my circuit using a BD5245-2M (though there is now BJT circuit, Q2&Q3, from the +5V to switch it in rather than just a diode, Im not sure what the designer has done this for), Im tempted to try this on the book rather than the max809 the designer has used, mind you the circuit diagrams do not say what threshold voltage max809 has been fitted, its possible its a bit low in an attempt to make it tolerate dips in the +5V??? Ill take a look at the markings when mine arrives
Did you spot if it as the PCB has R1 &C1 fitted ? as well as u14? the more I think about this the more likely I think it is that the designer was keeping his options open here...

There are a number of other small design improvements, the filter & series resistor on the 8051's ADC input monitoring the battery voltage now has series resistors, low pass filter (similar to my pocket386 mod) etc so it will give more constant readings, the higher battery voltage means he has had to include a potential divider, so the 8051s input is now better protected (this is what blew on my pocket386).

Yoghoo wrote on 2026-08-07, 22:19:

The USB-C to DC cable works indeed perfectly fine with the BOOK386. So can use an USB charger now.

Also designed a cover for the coprocessor/led gap. Used some clear translucent PETG so the LEDs are still visible. Ordered some darker translucent PETG so it fits the laptop better.

You read my mind! I thought a little cover would tidy things up nicely!
Did you put some vents in it? the power consumption of these coprocessors seems to vary quite a bit from one type to the next. Some are around 750mW, some as much as 2W (javispedro1 's +5V current reading upped by 400mA with his coprocessor fitted) I think enclosing them (certainly a 2W one) in a space that small may lead to overheating.. For ref the GC5420 that runs fairly warm in the pocket is 1.5W I think....
there is a nice summary of the co-processor options here, together with power consumption:
https://dougx.net/gaming/coproc.html

Could I have your design files so I can print one of my own?

Jules_nerd wrote on 2026-08-08, 18:21:

You read my mind! I thought a little cover would tidy things up nicely! Did you put some vents in it? the power consumption of […]
Show full quote

Yoghoo wrote on 2026-08-07, 22:19:

The USB-C to DC cable works indeed perfectly fine with the BOOK386. So can use an USB charger now.

Also designed a cover for the coprocessor/led gap. Used some clear translucent PETG so the LEDs are still visible. Ordered some darker translucent PETG so it fits the laptop better.

You read my mind! I thought a little cover would tidy things up nicely!
Did you put some vents in it? the power consumption of these coprocessors seems to vary quite a bit from one type to the next. Some are around 750mW, some as much as 2W (javispedro1 's +5V current reading upped by 400mA with his coprocessor fitted) I think enclosing them (certainly a 2W one) in a space that small may lead to overheating.. For ref the GC5420 that runs fairly warm in the pocket is 1.5W I think....
there is a nice summary of the co-processor options here, together with power consumption:
https://dougx.net/gaming/coproc.html

Could I have your design files so I can print one of my own?

I made 3 designs. One with some small holes (mainly for the leds), one with no holes and one that rest on the top cover. The first 2 'drop' tight inside the gap. I am waiting for the darker transparant PETG to decide which one I will use. Will take a couple of days because I have to dry it first as well. Will release it after that. See photo for the first and last one. Keep in mind that these were draft prints so they look a bit rough.

Looking at the prices for a suitable coprocessor I will never add one. So heat will not be a problem for me. I have enough desktops with added coprocessors and this laptop is just a toy and just to slow for anything slightly productive. Try starting Netscape in Windows 3.11 and you know what I mean. 😜

I suppose another possible would be to use black PLA or whatever for the bulk of it & light pipes made of PETG?
Ill put vents in one, I hear you, but I have a co-processor handy so....
A few posts ago when talking about win95I did say Id not tried any software that was processor intensive, then I rememberd Id tried netscape 😁 interestingly it ran web browsers slightly faster under win95, but i was using an ESP12 serial wifi module which really didn't help but it was clear the processor was struggling... esp when it came to rendering jpegs... 😁
It will be intresting to see how much a proper ethernet connection will speed things up.

I thought jpeg decoders used fixed point maths, but some say a co-processor speeds up jpeg rendering???

Jules_nerd wrote on 2026-08-08, 18:13:

Thanks for the summary, you put things in a rather more succinct readable form than I! ;-) Its a shame you cant quite fit more 1 […]
Show full quote

Thanks for the summary, you put things in a rather more succinct readable form than I! 😉
Its a shame you cant quite fit more 18650 s in, I was rather hoping an off the shelf 2s2p pack would go in! I mentioned the charge controller supports 3 cells, with a circuit mod, but a 2s2p pack (with inbuilt protection & ideally charge leveling), if the case can be modded, should work OK with the existing charge controller etc.

I was looking at the GD542* datasheets last night trying to work out if it was easy to do a swap like you mentioned, you've saved me some work there, thanks! Did you have to do a specific ROM for the GD5428 with 512kKB?
Is the GD5426 also a drop in replacement?

The coin cell will help of course, like when the main battery is flat... however on the pocket386 the lack of one was not the main issue causing low of date/time /BIOS resets.- I know, I tried putting one on ... and Ive had a 'scope hooked up to that supply to the M6117, it was rock solid when powering on/off/plugging in external supplies. Others on the net also report a battery did not help.
This is why i tried a proper PSU monitor IC to generate the PWG signal. I'm not sure why this isnt working 100% on the BOOK386, I can only assume its the extra delay i put in my circuit using a BD5245-2M (though there is now BJT circuit, Q2&Q3, from the +5V to switch it in rather than just a diode, Im not sure what the designer has done this for), Im tempted to try this on the book rather than the max809 the designer has used, mind you the circuit diagrams do not say what threshold voltage max809 has been fitted, its possible its a bit low in an attempt to make it tolerate dips in the +5V??? Ill take a look at the markings when mine arrives
Did you spot if it as the PCB has R1 &C1 fitted ? as well as u14? the more I think about this the more likely I think it is that the designer was keeping his options open here...

There are a number of other small design improvements, the filter & series resistor on the 8051's ADC input monitoring the battery voltage now has series resistors, low pass filter (similar to my pocket386 mod) etc so it will give more constant readings, the higher battery voltage means he has had to include a potential divider, so the 8051s input is now better protected (this is what blew on my pocket386).

Yeah you need to source a ROM that supports the GD542X in question (sadly none stretch all the way from GD5420 to GD5429 that I know of), and not to mention ensure the byte swapping is correct.

I didn't clock R1/C1/U14 and won't be re-opening the book386 this weekend, but can always check when I next do.

OK, I could prob program up an eprom.. what's special about the byte order on these?
this page seems to have a GD542x file.. not sure what they are getting at with the 'x'
https://theretroweb.com/chips/3840
The other issue is I dont think Ive a spare video card with one of these on.. so Id either be buying a complete card .. or taking my chances with a Chinese supplier...
where did you get yours from? any recommendations?

Yoghoo wrote on 2026-08-08, 12:28:

Anyone got the ps/2 mouse working on the BOOK386? Tested all ports (ps/2 keyboard, vga out, serial, parallel, game, usb, network) and all worked without problems. Serial mouse and touchpad are working but ps/2 mouse is not. Tried all combinations I could think off. First disabled only the touchpad (as described in the manual). Also tried disabling the mouse support in the bios and also disabling COM2. But none of the multiple mouse drivers I tried can find a ps/2 mouse.

Yes external PS/2 mouse worked for me on the first try. The manual says that "simultaneous touchpad and external PS/2 mouse are not supported under DOS/Windows", and suggests disabling the touchpad, but .... I have no clue why he would say that 😜 . Most tools that I know of will simply prefer the PS/2 one, or can be configured to prefer one of the two.

Which is very fortunate, because of course the 8501 is buggy and disabling the touchpad simply doesn't work 😐 . I can press FN+3, the OSD will show "TP OFF", and the touchpad will still happily send packets over COM2. So, if you really have a mouse driver that will prefer a mouse over serial rather than PS/2 , good luck ....
But at least MS Mouse prefers PS/2, as far as I can see.

By the way, keep "mouse support" in the BIOS definitely _on_ . It has nothing to do with the emulation -- it just changes the PS/2 BIOS to talk to the aux PS2 port. Most mouse drivers will require this to work with a PS/2 at all.
Note also PS/2 is not technically hotplug (i.e. you need to power off to disconnect/connect), just to eliminate common issues.

If you can't get one of the normal mouse drivers to work, may I suggest trying mine ? At least it will give you an error message coming from the BIOS that maybe we can try to understand... (and mine doesn't support serial at all, so it won't get confused by the touchpad).

javispedro1 wrote on 2026-08-08, 21:01:

If you can't get one of the normal mouse drivers to work, may I suggest trying mine ? At least it will give you an error message coming from the BIOS that maybe we can try to understand... (and mine doesn't support serial at all, so it won't get confused by the touchpad).

Will try your driver but your site is down atm. 😀 Tried not hot plugging, hot plugging, rebooting etc etc. But ps/2 mouse never worked for me.

Jules_nerd wrote on 2026-08-08, 19:40:

I suppose another possible would be to use black PLA or whatever for the bulk of it & light pipes made of PETG?

I thought jpeg decoders used fixed point maths, but some say a co-processor speeds up jpeg rendering???

Jpeg is slow as hell on these processors. Not something I want to try. 😜 Even my 486/66 is slow with those (depending on amount of colors of course). Netscape takes multiple minutes to load on the BOOK386 btw. And after that loading a site is no fun at all. Of course not something you would normally do on this CPU but I like to tinker with these things.

PLA and PETG don't bond together. So you need to glue it together and I hate gluing plastics as it most of the times leaves a haze or white residue. Even with those so called good glues.

I personally will use transparant dark PETG (without holes) or the one I showed earlier with the holes. The one with the holes looks very nice. Looks better IRL than on the photo. But I will release it in a couple of days so you can adjust it to your liking.

Heathen wrote on 2026-08-08, 14:16:

might be tempted to spin up an internal picogus

I would order one of those with fries, please 😀

More seriously, one of my complains is still all the emulated hardware from the 8051 or whatever new thing is on the book386. I can understand why some people may prefer the real OPL3 , but for all the other hardware, I don't understand the reason, since I think that by now we are emulating it better on a RaspberryPi than these random 3rd party clones...
A single RPi could probably handle GUS|SB + trackpad + keyboard + USB mice + ethernet|wifi and likely still have cycles to spare? And probably draw more useful information in the "new" display while doing it...

One of the small dreams I had is to report the charge of the battery to the x86 side ... which can't be done due to the closed 8051 fw. As far as features go, I know this one is pointless, but then one could add an APM BIOS and get the fancy "battery" icon in win95.

Yoghoo wrote on 2026-08-08, 21:22:

Will try your driver but your site is down atm. 😀 Tried not hot plugging, hot plugging, rebooting etc etc. But ps/2 mouse never worked for me.

Yeah, sorry, the git part is continously trounced by all these ridiculously hungry LLM bots trying to ingest all the source code in the planet, no matter how obsolete it is... but the files themselves I store on a static server and they resist a bit better.
https://depot.javispedro.com/vbox/vbados/rele … vbados_0_67.zip
https://depot.javispedro.com/vbox/vbados/rele … vbados_0_67.flp
The main site is also up again. EDIT: Nah it's down again. Better luck tomorrow.

javispedro1 wrote on 2026-08-08, 21:36:

Yeah, sorry, the git part is continously trounced by all these ridiculously hungry LLM bots trying to ingest all the source code […]
Show full quote

Yeah, sorry, the git part is continously trounced by all these ridiculously hungry LLM bots trying to ingest all the source code in the planet, no matter how obsolete it is... but the files themselves I store on a static server and they resist a bit better.
https://depot.javispedro.com/vbox/vbados/rele … vbados_0_67.zip
https://depot.javispedro.com/vbox/vbados/rele … vbados_0_67.flp
The main site is also up again.

Thanks. Downloaded it. Error running vbmouse.com:

Cannot init PS/2 mouse BIOS, err=4

After that the laptop hangs and needs to be restarted (ctrl-alt-del doesn't work). Tried with both touchpad enabled and disabled.

Yoghoo wrote on 2026-08-08, 21:48:

Cannot init PS/2 mouse BIOS, err=4

Well the BIOS call in question is https://www.ctyme.com/intr/rb-1596.htm#Table522, so 04h is "resend", which is what the BIOS responds when nothing is answering on the mouse side....
I get the same thing on my book386 when no mouse is connected.
(It should have nothing to do with "Mouse support" on the BIOS -- because if disabled I get a different error -- or the touchpad, which has no effect).
Does the mouse power up? You have another mouse to try?

javispedro1 wrote on 2026-08-08, 22:03:

Well the BIOS call in question is https://www.ctyme.com/intr/rb-1596.htm#Table522, so 04h is "resend", which is what the BIOS re […]
Show full quote

Yoghoo wrote on 2026-08-08, 21:48:

Cannot init PS/2 mouse BIOS, err=4

Well the BIOS call in question is https://www.ctyme.com/intr/rb-1596.htm#Table522, so 04h is "resend", which is what the BIOS responds when nothing is answering on the mouse side....
I get the same thing on my book386 when no mouse is connected.
(It should have nothing to do with "Mouse support" on the BIOS -- because if disabled I get a different error -- or the touchpad, which has no effect).
Does the mouse power up? You have another mouse to try?

Tried 5 different mouses (ball and optical). The optical mouses all get power (light is on). But nothing is detected. 🙁

Can somebody please make a screenshot/photo of the Advanced Chipset Setup screen from the BIOS? Especially someone with a working PS/2 mouse?

Yoghoo wrote on 2026-08-08, 22:41:

Can somebody please make a screensh
ot/photo of the Advanced Chipset Setup screen from the BIOS? Especially someone with a working PS/2 mouse?

EDIT: adding transcription of everything so that we have a searchable reference later on...

Which is basically the defaults. I don't think this is a BIOS setup issue. From my side I only have optical mice but they all work, even the wireless Intellimouse that didn't work on the pocket because it required two PS/2 plugs.
Just to check more obvious things, you're using the green port and leaving everything else unplugged, right?

javispedro1 wrote on 2026-08-08, 22:46:

BootUp Num-Lock Off BootUp Sequence C, A, CDROM Floppy Drive Swap Disabled Floppy Drive Seek Disabled Mouse Support Enabled Pass […]
Show full quote

Yoghoo wrote on 2026-08-08, 22:41:

Can somebody please make a screenshot/photo of the Advanced Chipset Setup screen from the BIOS? Especially someone with a working PS/2 mouse?

BootUp Num-Lock Off
BootUp Sequence C, A, CDROM
Floppy Drive Swap Disabled
Floppy Drive Seek Disabled
Mouse Support Enabled
Password Check Setup
C800, 32K Shadow Disabled
D000, 32K Shadow Disabled
D800, 32K Shadow Disabled
E000, 32K Shadow Disabled
E800, 32K Shadow Disabled

Which is basically the defaults. I don't think this is a BIOS setup issue. From my side I only have optical mice but they all work, even the wireless Intellimouse that didn't work on the pocket because it required two PS/2 plugs.
Just to check more obvious things, you're using the green port and leaving everything else unplugged, right?

Thats's the Advanced CMOS Setup. 😀 I need the Advanced Chipset Setup screen (GPCS, WatchDog etc).

Yes, I use the green port. Everything else is unplugged. Also like I said earlier all other ports work for example the PS/2 keyboard etc.

Yoghoo wrote on 2026-08-08, 22:52:

Thats's the Advanced CMOS Setup. 😀 I need the Advanced Chipset Setup screen (GPCS, WatchDog etc).

I have added all the pages, but I really don't think any of them can be related.
Sorry I can't think of more obvious things to try, I'd start looking at this as a hw issue.
I was going to suggest CuteMouse 2.0 which was the branch that supposedly skips the BIOS but it is definitely using the BIOS for detection too, so it'll fail there for sure (and anyway I still guess this is hw issue, not BIOS).
In theory it is possible to write a mouse driver that looks for a mouse in the keyboard PS/2 port. (or viceversa, a keyboard in mouse port). But to my knowledge no one has done this yet.

Jules_nerd wrote on 2026-08-08, 18:21:

Did you put some vents in it? the power consumption of these coprocessors seems to vary quite a bit from one type to the next. Some are around 750mW, some as much as 2W (javispedro1 's +5V current reading upped by 400mA with his coprocessor fitted) I think enclosing them (certainly a 2W one) in a space that small may lead to overheating.. F

The coproc definitely gets warm albeit fortunately it doesn't reach Book8086 levels of "warm". I suppose the relatively high power even at idle is because of the forced 40Mhz/sync mode which you mentioned and seems to be a m6117 limitation.
With the pwrcr trick the power consumption during HLT seems to be similar with and w/o coprocessor (I can reach down to 1A in both cases), so I think I'm going to leave it in.
I have no idea of the 386<->387 interface nor why this happens, but my random guess is most of the extra consumption I see is coming from the SoC, and not from the FPU itself.

javispedro1 wrote on 2026-08-08, 23:45:

I have added all the pages, but I really don't think any of them can be related. Sorry I can't think of more obvious things to […]
Show full quote

I have added all the pages, but I really don't think any of them can be related.
Sorry I can't think of more obvious things to try, I'd start looking at this as a hw issue.
I was going to suggest CuteMouse 2.0 which was the branch that supposedly skips the BIOS but it is definitely using the BIOS for detection too, so it'll fail there for sure (and anyway I still guess this is hw issue, not BIOS).
In theory it is possible to write a mouse driver that looks for a mouse in the keyboard PS/2 port. (or viceversa, a keyboard in mouse port). But to my knowledge no one has done this yet.

Thanks. I got the same settings so BIOS can scrapped than.

Will take a look at the internals. But that probably will have to wait until next weekend.

javispedro1 wrote on 2026-08-08, 23:45:

The coproc definitely gets warm albeit fortunately it doesn't reach Book8086 levels of "warm". I suppose the relatively high power even at idle is because of the forced 40Mhz/sync mode which you mentioned and seems to be a m6117 limitation.
With the pwrcr trick the power consumption during HLT seems to be similar with and w/o coprocessor (I can reach down to 1A in both cases), so I think I'm going to leave it in.
I have no idea of the 386<->387 interface nor why this happens, but my random guess is most of the extra consumption I see is coming from the SoC, and not from the FPU itself.

I quoted your 2W power consumption as an example of a real world result, but in the link I posted there are others that use something similar. It also seems that there is some variation form different examples of the same type....
and the need for vents in any cover? That's no more than my gut feeling....
Im not sure of the interface either, but i think the 40MHz clock for the co-processor is fror the M6117, so perhaps it disables this clock at the same time as its own internal one? when mine arrive i might stick a scope probe on it to find out!

Yoghoo wrote on 2026-08-09, 00:00:

javispedro1 wrote on 2026-08-08, 23:45:

I have added all the pages, but I really don't think any of them can be related. Sorry I can't think of more obvious things to […]
Show full quote

I have added all the pages, but I really don't think any of them can be related.
Sorry I can't think of more obvious things to try, I'd start looking at this as a hw issue.
I was going to suggest CuteMouse 2.0 which was the branch that supposedly skips the BIOS but it is definitely using the BIOS for detection too, so it'll fail there for sure (and anyway I still guess this is hw issue, not BIOS).
In theory it is possible to write a mouse driver that looks for a mouse in the keyboard PS/2 port. (or viceversa, a keyboard in mouse port). But to my knowledge no one has done this yet.

Thanks. I got the same settings so BIOS can scrapped than.

Will take a look at the internals. But that probably will have to wait until next weekend.

Im with others that this is kinda sounding hardware, unfortunately...
However, on the bios settings and serial port front, its handled by the M5113 (16550s in the pocket, on the option PCB), it's not hardware in the M6117, yet the BIOS has entries for 'OnBoard Serial Port' (same with parallel port really)
Is this just a name that's carried over from pinching the BIOS from another system and it doesn't matter? or does it have any other implications? its an area beyond me!
Edit,I did notice on the pocket, the parallel port settings in BIOS did nothing...., the address was fixed in hardware address decoding, though this was gated by a signal from the 8051(set from the enable parallel port option on the OSD) . Are these BIOS settings for this and the serial ports being passed on to the M5113???