bobd@hpuxa.ircc.ohio-state.edu (Bob DeBula) (03/14/89)
Last fall, I obtained an AT&T 6386 16 Mhz machine at a real reasonable (then) price. While I was installing my ESDI drive, I took time out to examine the upside-down motherboard. One peculiarity struck me immediately: the 386 chip is marked 20Mhz!!! Apparently AT&T (Olivetti) must have decided to standardize production. I also located the 32.000 Mhz clock chip (I think). Now the intriguing question is; Can I replace the clock chip (which appears to be soldered and I have no idea where to obtain the 40.000 Mhz equivalent) and get a 20 Mhz machine? My orignial memory which came with the machine was 80ns (overkill I think). I purchased 4 1Mb SIMMs from a chip peddler and installed them (and removed the 256K SIMMs which I promptly stuck in my MAC-II to make it a 2Mb machine). These SIMMs are 100ns (the chip peddler insisted that they work fine in a 20Mhz 6386). In any case, has anyone else noticed this (or done anything about it)? I think I'll hold off while the warranty lasts, but it seems awfully tempting to get a 20% or so increase in speed by replacing a clock crystal :-). Is there something I'm not thinking of (i.e. did Olivetti/AT&T put something in there to stop hardware hacks such as this?)? Thanks for any help you may give.
root@nebulus.UUCP (Dennis S. Breckenridge) (03/16/89)
In article <48@nisca.ircc.ohio-state.edu>, bobd@hpuxa.ircc.ohio-state.edu (Bob DeBula) writes: > > Last fall, I obtained an AT&T 6386 16 Mhz machine at a real reasonable (then) > price. While I was installing my ESDI drive, I took time out to examine > the upside-down motherboard. One peculiarity struck me immediately: the > 386 chip is marked 20Mhz!!! Apparently AT&T (Olivetti) must have decided > to standardize production. I also located the 32.000 Mhz clock chip (I > think). Now the intriguing question is; Can I replace the clock chip (which > appears to be soldered and I have no idea where to obtain the 40.000 Mhz > equivalent) and get a 20 Mhz machine? My orignial memory which came with the > machine was 80ns (overkill I think). I purchased 4 1Mb SIMMs from a chip > peddler and installed them (and removed the 256K SIMMs which I promptly > stuck in my MAC-II to make it a 2Mb machine). These SIMMs are 100ns (the > chip peddler insisted that they work fine in a 20Mhz 6386). In any case, > has anyone else noticed this (or done anything about it)? I think I'll > hold off while the warranty lasts, but it seems awfully tempting to get > a 20% or so increase in speed by replacing a clock crystal :-). Is there > something I'm not thinking of (i.e. did Olivetti/AT&T put something in there > to stop hardware hacks such as this?)? Thanks for any help you may give. Not all 6386 Machines have 20Mhz Chipsets, in fact I believe you lucked out. Comparing the 6386E the 6386 (20Mhz Desktop) and the 6386 (16mhz) I have found that not only the crystal changes but several of the the PAL's. One PAL I am sure changes is the PAL labeled PLYX or PLYS (depending on the run of the machine) This PAL is documented as the CYCLEGEN PAL in the operations manual. I have included a map of the 3 system cards with the PAL and PROM layouts. The ongoing Saga of the 6386-6386E-6386(20Mhz) as defined by PAL layouts. --------------------------- | CSL5 Tower/Desktop | | CSL2 20 Mhz | --------------------------- ---------- | PLYP | ---------- ------------------ Tower/Desktop | PBUQ (1.09) | 20 Mhz | PBVP (1.14) | ------------------ ---------- | PLYR | ---------- ------------------ Tower/Desktop | PBUZ (1.09) | 20 Mhz | PBVN (1.14) | ------------------ ---------- | PLYV | ---------- ---------- Tower/Desktop | PLYE | 20 Mhz | PLXX | ---------- ---------- ----------- Tower/Desktop/20 Mhz | PLYG | | PLYT | ---------- ----------- ---------- ----------- Desktop | PLYL | | PLYU | Tower/20 Mhz | PLYX | | | ---------- ----------- ---------- ----------- Desktop | PLYS | | PLY1 | Tower | PLYZ | | PLY2 | 20 Mhz | PLX7 | | PLXH | ---------- ----------- ---------- ----------- Tower/Desktop | PLY0 | | PLYK | 20 Mhz | PLXN | | PLXK | ---------- ----------- ---------- ----------- Desktop | PLYH | | PLX2 | Tower/20 mhz | PLYY | | | ---------- ----------- ------------------- | 80386 (16 Mhz) | | or | | 80386 (20 Mhz) | | | | | ------------------- I have uncovered some anomolies when experimenting with the various configurations. When the 20 Mhz System card is installed in the 6386E Tower The system behaves normally but the reverse is not true. The 6386E System card does not work at all in the 20 Mhz Desktop machine. When the CYCLEGEN 1.6 PAL is inserted in the 6386E the System sporatically RESETS itself (not a UNIX panic) but a hard reset. I summized that the CYCLGEN 1.6 is not a FAST device and the BUS timming is too far out of spec and the machine just resets. Every test I performed with the DESKTOP (20Mhz) machine performs flawlessly, including running the Computone AT-8 Serial card (with Driver Version 4.27) None of the BELL Technolgies cards work to my satisfaction (HUB-6, ACE and ICC) in fact they all panic the machine when UNIX 3.2 is using the modem (ttymXX) device. The Hard disk controller that is used in the Desktop (20Mhz) is a WD-1007-WAH with several components removed, nameley the HARD DISK BIOS eprom and the supporting logic. With the new SYSTEM BIOS (1.14) it now supports the 383Mb Hard disk under DOS as well as a new 81Mb Half Height ESDI CDC Drive. The problems with the Desktop are but one, It comes standard with a 3.5 inch floppy disk, DOS and UNIX come on 5.25 inch media. Even if we source a 5.25 inch drive the internal cable does not have the appropriate connector to match the drive in drive position 0. The second drive can be a 5.25 inch but you cannot boot the system from it. Dennis S. Breckenridge Integration Specialist National Technical Support AT&T Canada ============================================================================== "A mind is a terrible thing to MAIL: Dennis S. Breckenridge waste!" 206 Poyntz Ave North York, Ontario M2N1J6 (416) 733-1696 UUCP: uunet!attcan!nebulus!dennis ICBM: 43 4 58 N / 87 55 52 W 50 megatons should do! ==============================================================================
fmcgee@cuuxb.ATT.COM (Netnews Administrator) (03/16/89)
In article <48@nisca.ircc.ohio-state.edu> bobd@hpuxa.ircc.ohio-state.edu (Bob DeBula) writes: > >Last fall, I obtained an AT&T 6386 16 Mhz machine at a real reasonable (then) >price. While I was installing my ESDI drive, I took time out to examine >the upside-down motherboard. One peculiarity struck me immediately: the >386 chip is marked 20Mhz!!! Apparently AT&T (Olivetti) must have decided >to standardize production. I also located the 32.000 Mhz clock chip (I >think). Now the intriguing question is; Can I replace the clock chip (which >appears to be soldered and I have no idea where to obtain the 40.000 Mhz >equivalent) and get a 20 Mhz machine? My orignial memory which came with the >machine was 80ns (overkill I think). I purchased 4 1Mb SIMMs from a chip A few things.....The reason why the 20 Mhz. 386 is there is because Olivetti couldn't get any more 16 Mhz. chips, so they substituted 20 Mhz. cpu's. Since the board is designed to run at 16 Mhz., this doesn't buy you any performance improvement. I don't think you can just change the crystal for a number of reasons. The 20 Mhz. motherboards are not that similar; they have some different PALs as well. In addition, the 20 Mhz. bus board is not compatible with the 16 Mhz. bus board, so I don't think your motherboard will be able to talk to the busboard if you change the clock. The 80 ns. memory is also a similar situation to the 386 cpu's; they couldn't get 120's so the substituted 80's. Since the memory board is designed for 120's you don't pick up any performance by substituting 80's. Sorry, but doesn't look like you can upgrade to 20 Mhz. -- Frank McGee Tier 3 Indirect Channel Sales Support attmail!fmcgee
wtm@neoucom.UUCP (Bill Mayhew) (03/19/89)
In article <774@nebulus.UUCP>, root@nebulus.UUCP (Dennis S. Breckenridge) writes: > ... It comes standard with a 3.5 inch > floppy disk, DOS and UNIX come on 5.25 inch media. Even if we source a 5.25 > inch drive the internal cable does not have the appropriate connector to match > the drive in drive position 0. The second drive can be a 5.25 inch but you > cannot boot the system from it. > > Dennis S. Breckenridge Just for the recod, the last 6386 machine that we ordered arrived about a week ago. It was one of the package deal systems from the university grant program that comes with Unix + development set as a mandatory part of the deal. All the Unix goodies arrived on 1.44M 3-1/2 inch diskettes, which of course, was good since that is now the type of diskette drive all systems are shipped with. Just thought you'd want to know. I'm not sure if there is a different comm code for the 3-1/2 inch media distribution. If there is a demand, I'll try to find the purchase orders to see. --Bill wtm@impulse.UUCP The address above is more likely to result in me seeing your email than neoucom.UUCP.