Trying MSX SUPER TURBO in Msx2 HB-F9s...

Page 1/4
| 2 | 3 | 4

By Dhampird

Hero (585)

Dhampird's picture

28-07-2010, 03:56

Hello, i finished the build of the MSX SUPER TURBO circuit designed by Alwin Henseler, you can find the original design in this link: , I used the same components listed in the original site least some that I could not find and replaced by others that I quote below , the entire components that i used are:

ORIGINAL DESIGN ............................................... MY INSTALL

C1, C2, C3, C4 = 22 pF ......................................... The same
C5 = 330 pF ........................................................... I used 470pf 63V
D1...D6 ................................................................. I used 1N4148
IC1,IC2,IC3 ........................................................... The same
R1, R3 = 220 Ω ....................................................... The same
R2, R4 = 10 MΩ ..................................................... The same
R5 = 100 KΩ ........................................................... The same
R6 = 22 KΩ ............................................................. The same
R7 = 560 Ω ............................................................ The same
R8 = 47 Ω ............................................................... The same
T1 = BS170 ............................................................. The same
X1, X2 ..................................................................... 10Mhz
2 x 100 nF................................................................. I used 2 of 220nf each one
10 µF/10V.................................................................. I used 10uF/35V

The PCB finished:

I couldn´t test the circuit for the moment because i dont know very well how to install in my Msx2 Hb-F9s, and better will be to do the right things and no make mistakes with this i think. Also, i´m seeking to buy a most faster Z80 version, 8Mhz CMOS version.

It seems there is most ways to install, i´m just very confused with this, for now i made a squematic that shows some conexions that i dont know if will be correct or wrong, if anyone could help me to try the install in the hb-f9s will be great.


Login or register to post comments

By RetroTechie

Paragon (1563)

RetroTechie's picture

28-07-2010, 08:47

Can't believe you used original layout, board can be made much smaller with all SMD components (you can use small hole + pull through wire for via's). Anyway, the board looks fine - must have cost you some blood, sweat & tears... Wink

I suggest you hook up turbo circuit to +5V (outside of machine, external power supply). Then test if HC175 pin 4 responds (=changes logic level) to diode inputs being pulled low (hang a loose switch between any diode input & ground). If not, the FET may be reversed, or damaged by soldering heat / static discharge - both can easily happen with a small FET like this. Don't bother unsoldering & put in reversed - if this part of the circuit doesn't work, cut that FET out & replace with a new one. I usually bent it down on top of the diodes to reduce the vertical height.

A logic tester would be a great help, but high-impedance voltmeter should be sufficient.

Next test if that input signal is 'propagated' to HC175 outputs. You need both clock signals present for that, try (temporarily) hanging a 2nd crystal between the other X1 in/out points. If that works, you should see turbo LED output respond. I suggest you also put a LED on there temporarily for testing.

If HC175 pin 4 responds to diode inputs but turbo LED not, either the HC153 is broken (unlikely) or one or both crystal oscillator(s) aren't working. That would be a matter of checking connections & whether you used the correct component (values) in the right places. Again: logic tester would be a great help.

If that's fine the clock output will switch between X1 and X2 clock frequencies (keep schematic in mind for which is turbo frequency, and which is 3.58 MHz). Then you can hook it up to your MSX. Basically it takes the place of that red line you drew between VDP CPUCLK output & Z80 clock input (pin 6). So you'll have to cut that connection somewhere & 'inject' the turbo circuit's clock output there (and feed VDP's 3.58 MHz to X1 input of the turbo circuit). But check schematic to see how the CPU clock is routed, where it comes from & where it goes to (and check that on your mainboard, service manuals are often, but not always accurate). For one thing: it will also go to the MSX engine, and the engine will need the turbo'ed clock signal as well since it uses Z80 clock to time DRAM signals. If you connect a 3.58 MHz crystal between X1 points, you can leave that until you have everything fixed in place, and only then replace that crystal with VDP's 3.58 MHz signal.

That X1 diagram of yours shows RTC (Real Time Clock) circuit - it has nothing to do with CPU clock so forget that ASAP!!! (that's 32.768 KHz btw, not MHz). LOL!LOL! Normal Z80 clock is produced by VDP (21.47 MHz divided by 6).

Safe starter test would be to temporarily wire a diode input to ground (read: clock out goes to constant 3.58 MHz), and see if your machine still works normal. You can do this even if you didn't replace the Z80 yet. Then see if switching to a turbo setting works - start with a 'safe' / low speed like 6 Mhz. You do need to have Z80 replaced with faster version for this! Although Z80's often overclock well...

Probably screen output will be corrupted etc., but MSX should not hang I think (and that's what you'd be checking). If okay, wire up #IORQ to a diode input to fix VDP access / screen corruption.

After that it's a matter of finding a good place for turbo switch/LED, fix things in place etc. If LED is too bright, increase R7 to suite your taste. Make sure things stay where you want them, also with machine upside down, shaken or after many years. For example double-sided sticky may not be good enough as it may come loose after time. When the add-on board goes moving, a short circuit is easily made -> poof, dead MSX. Crying

Then find max speed, and seeing what other components need switch-back, or replacement with faster version. There's no floppy drive in the HB-F9S, so that's easy, but slow EPROMs might be a problem, what speed are they? FWIW: you'd be very lucky if that 10 MHz works at all (and reliable) - start with something like 6/7/8 Mhz (read: have a selection of crystals at hand). Best way to test is with regularly used peripherals plugged in, run different software / mapper test programs etc.

As you see, you can do this in many small steps, so you see immediately what's the problem if something doesn't work. I would go through such an upgrade much faster, but suggest you go the small-steps route, it'll get you results faster believe me. Keep us posted with your progress? (if you takes notes of the exact changes you make in the machine, you'll be left with instructions that others can follow).

By RetroTechie

Paragon (1563)

RetroTechie's picture

28-07-2010, 11:07

Just spotted you have + and - reversed on the LED in your picture. Smile2

For that switch, you don't have to connect it to ground on the turbo circuit itself (maybe just for testing). For example you can pick up a ground point very near where you install the switch, and only route the other side over the board to the turbo circuit. Same thing for LED signal. Like so: (neither are timing-critical so it's okay to use long wires if you want, pull a 2-wire strand from a flatcable & it looks even better)

LED(+)  --------+
LED(-)  ---+    |
           |    +------------------------      /  /     ----------<----------  turbo LED output
           |    +------------------------    /   /      ------------------->  diode input (or common side of diodes)
Switch ---+     |

(aargh... no way to have monospaced font here? )
Some people like switch with integrated LED, on my Sanyo Wavy I have a 'click in, click out' push-button that only required a small, round hole to be drilled. Smile

By Dhampird

Hero (585)

Dhampird's picture

30-07-2010, 06:03

Hello RetroTechie, thanks for the fully info and reply, concerning to the tests mentioned, I tried to do the test paying attention to what you explained, i dont know if all that i did is the same that you expains to test circuit, first i tried TEST1 (test if HC175 pin 4 responds), i checked continuity and voltage, and i get some values as explained in the images:

After that i put the diode and another 10mhz cristal in X1 points, diode work, but also work if i dont put the cristal in X1, the images are:

It seems that the circuit work, but i do not know if I need to test something more than what is shown in the images.

Ok, for the conections to the msx mainboard i follow the tracks: "pin 6 Z80 CLK" and "pin 8 VDP cpuclock", using the hb-f9s service manual and also in the physical msx mainboard, it seems that the tracks do not join together but go to the same IC 24 to different pins, in service manual the IC24 knows how "RAM CONTROL SIGNAL GEN", for now i´m just trying to find a way and see carefully the service manual, in the midst of these each tracks is a bridge that could be cut fro the install, for the moment i dont know if will be the correct point to install the circuit. I paste two images with the marked tracks, and i think the finally correct conections to do in msx super turbo.

About the Z80, i put a socket on it, i changed all the RAM 512K install and i used the IC13 socket for it to release as much old welding in z80, about the new Z80 i´m seeking a 8Mhz CMOS version, i have for now 10Mhz and 14Mhz cristals, about the eproms i changed it to fast versions updated it to Msx2+, for IC11 i used 27c256 (70ns) and for IC12 i used 27c512 (90ns) ,this 27c512 possible will change for another with 80ns ( i couldn´t find the 27c512 more faster than 80 ns).

No problem about to post in the forum the results as long as this in my hand make the correct tests. No problem with that.

By RetroTechie

Paragon (1563)

RetroTechie's picture

30-07-2010, 11:50

You're confusing 'on' & 'off' state of your switch: Hannibal on = connection = diode input grounded -> 3.58 MHz, off = no connection = diode input open -> turbo. Anyway: measurements look okay, try putting black pin to circuit ground next time, makes for easier reading...

It seems that the circuit work, but i do not know if I need to test something more than what is shown in the images.
Nothing more to test I suppose, unless you have fancy stuff like oscilloscope or logic analyzer (neither of which I have). Above pix are a bit overkill, but they are very helpful.

I checked the clock routing myself in service manual, and compared with your last pic:

  • VDP clock output (shown as "CLK" in service manual) first goes to pull-up resistor R169, IC27 pin 5 and IC28 pin 7 (composite video encoding, on the right in your pic). This should be left in place, and this is the 3.58 MHz signal you should connect to X1 IN point on the turbo circuit. Yellow marked wire bridge, or some point near R169 looks like good spot to pick it up. Or use one of IC24 output pins 10 / 12.
  • Then it continues on to IC24 pin 11. Not needed anymore, but you can leave IC/input in place. Corresponding output pin 10 (marked "CPUCLK" in schematic) goes on to S1985 pin 30 and Z80 pin 6. This output should be isolated, and S1985/Z80 connected to clock output of the turbo circuit. I also see a pull-up resistor on Z80 clock input (pin 6), it's probably okay to remove that when turbo circuit is hooked up.
  • VDP's 3.58 MHz also goes to IC 24 pin 13. Corresponding output pin 12 (marked "CLOCK" in schematic) can't be traced in your pic, but according to schematic, goes to cartridge slot pins 42. This output should also be isolated, and those cartridge slot pins 42 connected to turbo circuit's clock output as well.

It looks like that last signal also goes over a wire bridge, between those you marked red & yellow. In that case, I'd do something like:

  • Cut that 'red' wire bridge, so that IC24 output pin 10 goes nowhere and S1985 pin 30 / Z80 pin 6 are 'hanging in the air'.
  • Cut that middle wire bridge (between red & yellow), so that IC24 pin 12 goes nowhere and cartridge slot pins 42 are hanging in the air.
  • Solder the upper ends of those wire bridges together, so that S1985 pin 30, Z80 pin 6 and cartridgeslot pins 42 become 1 electrical point. Then connect that to turbo circuit's clock output.

Seems like the easiest way to re-route the clock signal here... Btw: don't trust me blindly on this one! Check all those connections on your board to make sure how they run! When you've made necessary connections: check everything again. Before you're ready to power it up, check a 3rd time. It's unlikely you'll do irreversible damage when you make a mistake re-routing clock signals, but I suppose you'll understand the work you'll have on your hands when eg. VDP's or turbo circuit's clock output is damaged... Crazy

For the rest: get a more realistic set of quartz crystals! Suggestion: 3.58, 7.3 & 8 MHz. 90 ns EPROMs should be fast enough to go up to ~10 MHz, but limits are elsewhere: peripherals, S1985 engine internal delays, DRAM timing, circuit board trace lengths (not from speed of electric signals, but because of signal distortion), etc, etc. And max speed reachable != max speed stable.

By Dhampird

Hero (585)

Dhampird's picture

04-08-2010, 04:09

Hello RetroTechie finally i hooked up the Msx Super Turbo circuit to the Msx, for the moment i dont have the new Z80, i´m waiting arrive for an 8Mhz Cmos version that i bought in ebay, i dont have any 3.58, 7.3 & 8 MHz cristals (tomorrow i think i could buy them), i have installed in the circuit 10Mhz cristal , i have the original Z80 in the msx, for now i installed the circuit in test mode, i wired a Input Diode to Ground (clock out goes to constant 3.58 MHz), also i removed the z80 resistor you told me, and i made the wire to z80 pin 20 to a diode input to fix VDP access / screen corruption.

I turned on the msx and worked perfectly, i tried many times and worked normaly.

The diode in this mode not work, i conected the - diode leg and also one leg of the switch to GND, the + Diode leg to Turbo led conection, and the other leg switch to a diode input, not diode light in On position of the switch, i think the way that i hooked up is right, please tell me if you know the problem or if in this test mode the diode is normal that not working.

I´m wonder if i can do more kind of test for now, if i dont have any more cristals or the new z80.

I think the next step would be procedure (tell me if i wrong please):

-Change the 10mhz cristal for another one like 6Mhz or less,
-Change the Z80 for the new 8mhz z80
-Remove diode input to ground conection and use this input for the switch.

Ok, here are the pictures of the install:

Thanks for all the help, again.

By RetroTechie

Paragon (1563)

RetroTechie's picture

04-08-2010, 15:23

Nice & short wiring, you're learning... Smile2 How did you (mechanically) fix the add-on board in place?

The diode in this mode not work, i conected the - diode leg and also one leg of the switch to GND, the + Diode leg to Turbo led conection, and the other leg switch to a diode input, not diode light in On position of the switch, i think the way that i hooked up is right, please tell me if you know the problem or if in this test mode the diode is normal that not working.
Didn't you read circuit description Question If any diode input is low, circuit switches to X1 frequency (3.58 MHz) & turbo LED goes off. With that switch connected, you might as well remove diode input -> ground wire, switch on has same effect. Likewise, the Z80 #IORQ -> diode input has no effect @3.58 MHz.

Btw. be careful where that switch is whenever you power on the board! If you forget & put the board on top of it, you can (will !) have short circuit wherever switch touches that board. Touch unlucky point & kill your MSX... Crying (I managed to do something similar with a ZX Spectrum once, cost me a ZX Spectrum ULA chip).

I´m wonder if i can do more kind of test for now, if i dont have any more cristals or the new z80.
Well you could place a socket for that Z80 - oh wait, you already did :-)

Also check peripherals that use clock signal in both cartridge slots: DRAM based memory mappers, sound carts like SCC / FM-PAC etc. When turbo is working, you should re-do that check, and you'll find that well built memory mappers work fine at least till ~7 MHz, not so well built ones will fail at turbo speed. Sound carts will go up in pitch, which is normal for turbo-upgraded machines (there are several ways to fix that, or you just switch back to normal speed whenever you use those).

Only other thing I can think of, is check some junk electronics to see if you can find some 4.?? MHz crystal (low enough that normal Z80 will work at that speed, but different from 3.58 MHz enough for you to do some tests).

By Dhampird

Hero (585)

Dhampird's picture

05-08-2010, 03:31

Hello again, for fix the add-on board in place i used thick double-sided tape, the Msx Super Turbo dont move with this, but when i finished all the install i will put some thermal silicon in the corners for more security and so avoid that the double-sided tape dont release it.

Thanks again for the explanation of how Turbo Led works, and also for notice about the switch, i´ll try to change the wires and use more long cables for avoid problems with it.

For now, i test some carts, the MMC/SD sinfox and the LPE-SAN3-V2 sound cart, and no problems for the moment, but i dont release for now the diode to ground.( i have only problems with the RAM, but it not has to do with the msx turbo, is a older problem, it seems that sometimes in boot i see 256K and not 512K, with msx2+ roms that i install now i can see all memory in boot and sometimes show up 256K and sometimes 512K, it seems that if i use testram and the boot up shows 256K the testram detects only 256K, and if the boot up shows 512K the testram detects 512K, i think is a problem in X5 circuit, i posted the problem in my RAM post).

I´ll try to find the cristals and a 4Mhz cristal, to try, the new z80 is on the way and i hopefully that will come soon.
I hope to have something new soon.

Thanks as always for all your help.

By Dhampird

Hero (585)

Dhampird's picture

08-08-2010, 23:21

Hello, finally I changed the 10 MHz crystal with one of 4MHz, I removed "Diode to Ground" and "Z80 #IORQ to Diode" connections and i tried some test, for now I have these tests done on the original z80 hb-f9s, the MSX SUPER TURBO working!

I tested some games with SWITCH position OFF = TURBO mode ON, for the test i loaded some .roms in LPE-SAN3-V2 , this card has in built a stereo output, in some games the music sounds different tones when i set the TURBO to ON, but dont be affected to the stereo output of this LPE-SAN3-V2 sound cartidge.
Also comment that i get some strange symbols in Msxdos2 if i use the TURBO ON on it.
For now i tested this games:

King's Valley 2 in 4mhz.___________working more faster and no sound problems, in this game the character moves faster on screen when TURBO is ON (led ON)
Vampire killer in 4mhz____________ working more faster and the sound is different now, in this game sometimes the character moves faster on screen when TURBO is ON, the sound changes some tune, is diferrent or sound like off-key instrument.
Metal Gear 1____________________same effect that in vampire killer happens.
Nemesis 2______________________no changes from normal game, in speed or sound remains the same.
F1 Spirit________________________no changes from normal game, in speed or sound remains the same.
Parodius________________________no changes from normal game, in speed or sound remains the same.
Contra__________________________no changes from normal game, in speed or sound remains the same.

I´m waiting the arrive of new z80 8Mhz CMOS, new SRAM (real static, not pseudo) memory and more crystals.

If will be possible more test with the parts that i have now I would be happy to try.

By Dhampird

Hero (585)

Dhampird's picture

25-08-2010, 21:31

Hello, it seems that I have solved the problem with the 512K SRAM, I've done some testing and now always shows in the boot 512K, due to this problem I could not update the comments about the tests I've been doing with the Msx Super Turbo . Now i´ll continue commenting the experience with the Msx Super Turbo.

Ok, about the last post where i told about strange symbols in Msxdos2 and Basic if the Turbo was in ON, RetroTechie told me about this:

"Try reconnecting Z80 #IORQ signal to one of the diode inputs. You can use VDP #CSR and #CSR signals instead to not switch back for *every* I/O port, but then you'd also need wires to switch back for MSX-Music, MSX-Audio etc, and those signals are not inside the machine but inside cartridges. In other words: much troubles for little gain. Using #IORQ to switch back to 3.58 Mhz is simple & works well. Many peripherals have been tested @ 7 MHz, but beyond that there's no telling what will work or not. And HB-F9P/S doesn't have much signal buffers between insides & cartridge slots, so *number* of cartridges, or # of connections on individual signals may matter a lot. In other words: test many combinations before claiming that a cartridge works at some speed..."

After reconnecting Z80 #IORQ signal to one of the diode inputs, the strange problem symbols solved.

Here you have some sample images relating to the problem of strange Symbols in MsxDos 2:

After solve the problem, finally the new Z80 arrived: Z80 8MHZ CPU Z84C0008PEG, that i I placed in the socket pre-installed, i tested and worked with no problem. Then i placed again the original 3.58 MHz Z80 to the socket to try the following TESTS:

-Overclock the original 3.58 MHz Z80 to 6Mhz, putting a 6mhz crystal instead 4mhz in Msx Super Turbo. My original Z80 3.58Mhz is labeled like: NEC Japan 8616M5 / D780C -1, the result of this test is that some games have problems with sound, but the speed increase clearly shows now in some games, I uploaded a video on youtube in which i test Metal Gear 2, in this game the sound is normal and you´ll see the speedup of Solid Snake in his movements, also tell that when the colonel calls from codec, the call advice appear in Japanese but should appear in Spanish (my MG2 is Spanish version) and i saw some glitches in screen that sometimes solves if i set to TURBO OFF, sometimes solves and sometimes the glitches remain even when the Turbo is OFF. I didnt try much games in this test, but if needed no problem to try more tests.
6MHz probably say that is too much for the Z80 original version NEC Japan 8616M5 / D780C -1

"Msx Super Turbo with 6Mhz Cristal in HB-F9s Original Z80 3.58Mhz"

-After test the 6mhz in the original HB-F9s Z80, i tested the 6Mhz with the new 8MHZ Z80 CPU Z84C0008PEG, the results have been perfect in MG2, no problem glitches, sound OK, and speed up ok. Anyways in another games persist the sound problems. You could see the results in another video, note that if you look when solid snake crawling on the floor (sneaking) you´ll see the increase of speed better.

"Msx Super Turbo with 6Mhz Cristal in HB-F9s NEW 8MHZ Z80 CPU Z84C0008PEG"

Another game tested with the 6Mhz and Z80 8Mhz was Nemesis2, sometimes the game increases the speed, but the increase of speed is not constant, the sound also is ok.
Another tested was Vampire Killer and the speed is no constant, the sound suffer distorsion on it.

Note that if you see carefully at the right below part of the screen in the videos, you´ll see the red light (Turbo ON) flashing, this is normal and RetroTechie explains me about it:

"Flashing turbo LED is normal: it means turbo is on, but regularly switched back to 3.58 MHz (on/off ratio gives an indication of average speed). With only #IORQ connected to diode inputs, that means: frequent access to I/O ports. In practice that means: intensive VDP acccess (like scrolling text, during action games etc). For example a memory mapper-test may show same effect since block switching = writes to FC-FFh I/O ports."

In all this tests that i made when i refered to the sound i´m talking about the msx sound not the LPE-SANS sound cartidge sound output, the cart LPE-SANS sound output always remains stable as explained me Retrotechie because this cartridge has a own crystal built on it.

-Now, related to RetroTechie told me by mail i like to try another way to install the Msx Super Turbo, RetroTechie explained me by mail:

"Also interesting would be if machine works with fixed 3.58 MHz (instead of CPU clock) fed to S1985, to have normal PSG sound @turbo. Will not work since you put DRAMs back in, but may be fine when main memory = SRAM. Maybe you'd like to try that later... I meant specifically *the S1985 MSX engine* clock input. That is: disconnect S1985 clock pin from that common point (soldered together wire bridges) and connect it to fixed 3.58 MHz signal (=what you picked up as X1 input signal), but leave Z80 & cartridge slot pins connected to turbo circuit clock output (& no changes on turbo circuit). Again: *will not* work while using DRAMs, may work when using SRAM for main memory. More useful to try when you have your MSX at a higher speed than now (refered to the old 4Mhz speed)"

Ok, before changing anything of the Msx Super Turbo install, I made a kind of diagram that shows correctly (if I'm right) the explanation RetroTechie told me to do this new mod to get normal PSG sound in Turbo, here you are:

I think that this diagram show the correct install, but i prefer a confirmation that it is correct to prevent any damage in the Msx.

Well, i´m try to get some 7Mhz and 9Mhz crystals, i dont find this crystals in the normal record stores but i think that i will have to do a deep search to get it...
I have 8Mhz crystals, i could to try these.

Regards and thanks again Retrotechie for the fully help, when I get some new results i´ll post again.

By RetroTechie

Paragon (1563)

RetroTechie's picture

29-08-2010, 07:30

Nope... Tongue in the above picture you'd feed Z80 a different signal (3.58) than the cartridge slots (turbo) - no good, they all have to keep the same signal.

Follow that horizontal circuit track coming from S1985 pin 30 - a little bit further it gets to a 'T-split', up goes to Z80, right goes to soldered together wire bridges. To feed 3,58 MHz to only the S1985, you'd have to cut that short track between S1985 and 'T-split' (so that S1985 pin 30 is the only pin that gets disconnected). And then connect that S1985 pin 30 to VDP's 3.58 MHz output (the long yellow marked track in above picture). Ignore all other connections: you'd be leaving everything else as is.

Good place for a cut would be just left of the point where the track starts to corner (so that wire to left part is short, and track is easy to repair if mod doesn't work). Whether you want to try this, is up to you & I really don't know if it will work. But it's easy to try, would be nice if it works, and good to know whether it does (either way). Result should basically tell if S1985 uses the clock signal for anything else besides DRAM timing and PSG (I'm guessing not, but not sure).

As for crystal frequencies: the Wiki has a long list, but only small number of these are common. In our area of interest you're probably looking at 7.16 - 7.37 - 8.00 - 8.87 - 9.22 or 9.83 MHz to choose from. Not all shops have all values, but none of these are hard to find (perhaps from junk electronics?). Try the bigger housing if that's available & mini housing not. FWIW: 7.37 Mhz should be safe, perhaps 8.00 MHz too but don't be surprised if some peripherals don't work then, at >8 MHz expect machine stability issues & problems with many peripherals.

Page 1/4
| 2 | 3 | 4