• Please review our updated Terms and Rules here

Tandy 1000 SL overclocking

The gglabs CGA2RGB is.

The Samsung 710N VGA monitor is alive and well but can't unfortunately auto adjust the image to a satisfactory level, and its image isn't as sharp as my LCD TV via the SCART conector.

The Tandy motherboard just needs new and faster RAM, a separate oscillator and possibly a replacement serial EEPROM.
 
Trixter said:
You're not out $35. My gglabs unit was initially damaged, and gglabs replaced my unit for free. Contact him.
Click to expand...

Major pain in the butt, since the unit wasn't sent directly to me but via Shipito forwarding service, as it is just cheaper than sending via USPS overseas. But yeah, i'll contact him and see what can be done. My best guess is that the PAL is screwed.
 
CarlosTex said:
I think i might have to replace the serial EEPROM. Every now and then feels like settings are somehow changed between power cycles. This was driving me nuts, leading to a point where a fixed disk couldn't be booted and DOS in ROM would be default again. I had to remove the Zilog floppy disk controller so that it would force serching for fixed disks so i could run SETUPSL. When i ran SETUPSL in advanced mode i noticed that RAM was set for 3 waitstates, which explains why the system was so slow in memory operations when i benchmarked it.
Click to expand...

Side note on this... do we know which bits control what in the serial EEPROM? I don't remember that being in the tech manual. Just curious.
 
Trixter said:
Assuming you've tried both pins 1-2 and 2-3 for sync generation: 1-2 is 2V composite Sync (required for GBS-8200) and 2-3 is 5V separate H-Sync and V-Sync for 15KHz capable monitors.
Click to expand...

Yes i did try that, but the adapter seems to be faulty.

I got in contact with the seller, he even sent me a .jed file so i could progra a different GAL but the problem persists.



As far as the final objective, overclocking the CPU to 10MHz: it is possible, the system does boot with a 30MHz oscillator. I get to a DOS prompt without a problem, but things go downhill from there. Whatever i try to execute just ends in errors. Even a dir command results in a mess of characters and system lockup. I don't think the RAM i have can handle the speed, most DRAM on the board is rated for 150ns, although there are some 120ns in there. It seems that 10MHz is just too much for the RAM. This also explains why at 9.54Mhz the system was unstable, with games crashing and the common video artifacts.

I already ordered a whole bunch of NEC branded 41464 DRAM rated for 100ns. I got enough to replace all the RAM, including the video RAM too. The RAM will probably take 5 to 6 weeks to arrive, so its going to be a looong wait for me. Still with the 100ns RAM i'm not 100% sure it will work without any issues, although i seem to remember that 10MHz Tandy machines, like the Tandy TL/3 do indeed use memory rated for 100ns.

If the system is not stable enough at 10MHz even with the 100ns RAM, it will most certainly be if i use a 28.6MHz oscillator. It will only give me a 9.54Mhz CPU clock, but i'll be happy enough with that boost.
 
Update:

-gglabs unit is definitely dead,contacted the seller and waiting for a response;

I cannibalized around 20 HY53C464LS-80 from VGA cards (the Hakko FR-300 is an AMAZING desoldering gun), removed all the old DRAM and replaced them with the cannibalized parts. Supposedly these parts are 80ns, so more than fast enough for what i wanted.
But result is pretty much the same as before at 10MHz.
So i tried tapping the 28.6 crystal on the board again for 9.54MHz speed and now the system worked fine. Games, programs everything working just great. So i tried to get the system faster by running SETUPSL /A for 0 waistate memory. I saved the settings, rebooted, ran checkit and...weird graphic corruption again! WTF? So i ran SETUPSL /A again and restored BIOS defaults. I was sure that this would solve the issue but it didn't. My guess is that that serial EEPROM surely doesn't hold data very well now. I checked the datasheet and it states that it should be good for 40,000 write cycles and 10 year data retention. This is surely not a very high endurance serial EEPROM for sure, and it seems that if i write to it when the system is running at a speed of anything greater than 8MHz it just craps out. I ordered a pin compatible ATMEL 93C46E serial EEPROM to replace the other.

I also took a look into the datasheet as well, and noticed that besides the CPU, 2 other IC's will take the CPUCLK generated by the buffer blue. One of them is a 74 series logic chip a 74LS175 flip flop and a PAL16R4A. From what i could understand from the datasheet, the PAL16R4A can take clock speeds up tto 25MHz so that shouldn't be an issue. The 74LS series chips are low power versions of the 74S versions which were in other way very fast but with high power consumption. The LS are exactly the opposite, in fact the same speed as the original 74 series released in 1966. So i ordered a 74ALS175 flip flop to replace the old one although i have no idea if this is the main issue.

Some good news otherwise is that the case keenerb sent me is now in my hands. Spent a couple hours at customs to pick it up, but i'm happy and surprised how small it is:


DG9_nvMWsAAjxdK.jpg:large




In any case i can always try to find a TL motherboard which i was told fits the SL case. The TL has more or less the speed i would like.
 
Ok, i finally had some success with this.

I'm now using a 27Mhz oscillator and the system seems to be rock stable. Tried several programs, several games and everything seems to run fine. A very modest speed improvement but still noticeable. I don't think DMA can take much more of this. Even though the datasheet tell the DMA is equivalent to the 8237A-5 (which should be enough to accept a 30MHz OSCI) 10MHz clock operation seems to much, and the system just craps out. It could be this and/or the 74 series logic need some replacements, the LS versions are low power but not exactly speedy.

Here are some pics:

DHl94DvXcAAccPC.jpg


DHl95ZfXgAAIKbr.jpg



I still think that NEC V30 CPU's have a huge potential that i've never seen unleashed. I suspect a V30HL running at 16MHz would be very fast, probably faster than a 8MHz IBM AT and it would be faster than a TL/3, making this a very unique system. Only way this is possible is to build a Turbo board that drives the CPU independently from the other motherboard clocks. The Tandy 1000 SL seems to use all clocks with a 33% clock duty cycle. I don't know much of electronics, so i don't know if building such a circuit is a trivial task or not.
 
Trixter said:
Nice, you've just about managed 286 performance out of that system.
Click to expand...


Yeah, should be close to the first revision IBM AT @6 Mhz. Too bad i can't get the full 10MHz, and its really a shame that no one can use a NEC V30HL to its full 16Mhz potential. I often wonder how simple/complicated would be to design an independent clock circuit just for the CPU avoiding messing with other clocks on the system. A Tandy 1000 SL with such speed flexibility would allow for a very unique system, considering that SL/TL series can be used with a VGA card and switch Tandy/VGA via VSWITCH. High Density Drives are also easy to add, plus a Dallas 1216E RTC module and an AdLib card and you would get an extremely flexible system that could cover a wide range of games for mid to late 1980's. A 16MHz V30 would be very competent and close to entry level 286 clones, i'm thinking 10MHz 286's. Maybe the V30HL would be tolerant of a slighty higher clockspeed too, like 17 or 18MHz without crapping out.

It would definitely be among the fastest XT class PC's EVER, without using 286/386/486 accelerator boards.
 
CarlosTex said:
It would definitely be among the fastest XT class PC's EVER, without using 286/386/486 accelerator boards.
Click to expand...

What about an 8086 accelerator board? I obtained an AST FlashPak which is an 8086/NEC V30 replacement board. Why would this be faster? It comes with 8K L2 cache. :-O They claim a 3x speedup, which of course I'll be able to verify with TOPBENCH.
 
After 9 months of procrastination I finally got the last piece of my 1000 SL completed. I'd previously installed an 8087 copper and battery clock. The software is finally installed! It took TWO TWEENERS and two thumb drives to finally transfer the program and then used GEOS to edit Autoexec.bat. It boots to initially to 1917 date but shows 2017 everywhere else. I'm elated! Now to find an EMS card and attempt a Windows 3.1 install. Will the previously installed V30 let it run?
 
Rick Ethridge said:
I'm elated! Now to find an EMS card and attempt a Windows 3.1 install. Will the previously installed V30 let it run?
Click to expand...
The SL lacks an a20 gate, meaning Windows 3.0a (3.00a) is as high as you can go.

I unlisted the demonstration video I'd made, but I find Windows 3.0 to be a very functional, usable OS with this system. EMS is definitely recommended.
 
Ok tried to replicate your mod CarlosTex on my SL, cut the trace going to R24.
Jury Rigged a small protoboard soldering a socket fro the can crystals and just connected gnd and +5 using a molex connector and the osc output to the R24 resistor on the motherboard.

Now i have the next CPUs to test
Ceramic NEC V30 D701160-8 (the slow one)
Plastic NEC V30 D70116HCZ-10 (this is the 10mhz one)

I have only a few can osc.
19mhz (works slow as expected)
24mhz (works but same as stock got 2.97 time IBM PC-XT on checkit)
30mhz (don't post black screen)
50mhz (don't post black screen)

So its a partial success the system at least worked.
May i ask CarlosTex where you purchased your 16mhz V30 ?.
And also i believe you started using a 10mhz v30 also ?, what speed you where able to get from it and using what can osc speeds?.
I have read the threads several times but its easy to get lost!.

Some pictures:
 
Hi, sorry i haven't been around for a while.

You can get a V30 HL on eBay, make sure you just order the right one. Sometimes the picture is not correct but you just need to make sure part number is correct, D70116HCZ-16 is the one you want.

You can also find them at utsource i believe. A 48MHz oscillator will do the trick.
 
Finally got a D70116HCZ-16, but no luck the computer will work using a 24mhz crystal or slower ones of course slowing down according to the speed of the crystal so i think i got the mod right.
But if i use any crystal faster than 24mhz i got a black screen no post.
Not sure why i don't work for me the modification is super simple :confused:
 
Hi....It seems that the 24MHz clock goes into the IC 8079024 (U41), chip which generates the CPU CLK and others. This might now be a good idea, but it would still be fun to try it out. It will probably muck up the locks for other stuff.
 
You must log in or register to reply here.
Back
Top