retro 8088 project

[pb]

I'm working on a (what I thought would be) simple 8088 based retro thing.  The Fujitsu 8088-2 that came from digikey doesn't have a notch or single dot index on it.  Instead it seems to have a groove/index along one side of the chip.  Based on the orientation and similarities to the ceramic package version in the datasheet, I believe that the indexed side marks pins 1-20.  Is that the correct interpretation?  Unless I'm missing something, I can't find explicit confirmation in the datasheet.

The reason I am asking is that the circuit (based on the 8284/8088 circuit in Grossblatt's 8088 Project Book) is behaving oddly.  CLOCK and RESET out of the 8284 look correct but the address and control lines out of the 8088 don't seem to change from all zeros (not even tri-stated) after RESET goes back low.  There is some noise/bouncing on ALE and #RD when RESET first goes high but nothing after RESET returns low.  What I was expecting to see is the boot address and ALE to signal the initial instruction read.  I've tried measuring ALE using a scope and a LA.  Neither seemed to pick up any change in ALE (always zero).  There is no memory (RAM/ROM) connected yet so maybe my expectations are incorrect.  I'm running out of hw debugging skill.  Any help would be appreciated.

thanks...

[pb]

no thoughts from the masses?

[Feanor]

Likely no-one can find the specific data sheet you are looking at there are w hole lot of Intel ones out there all with appropriate notches. How about some links, pictures and screen shots?

[slburris]

How are you connecting the -RST and -AEN1, -AEN2, RDY1, and RDY2 inputs to the 8284?
If they are unconnected, the processor could be held in reset or be waiting
eternally for a bus cycle to complete.

Look at the RESET and READY pins on the processor.  RESET show be low and READY should be
high.

You could also put a scope on the processor's clock input and make sure it looks
OK.  The old processors were pretty particular about voltage levels, i..e.
they weren't TTL compatible.  Hence the need for the clock generator chip.

Scott

[pb]

How are you connecting the -RST and -AEN1, -AEN2, RDY1, and RDY2 inputs to the 8284?
If they are unconnected, the processor could be held in reset or be waiting
eternally for a bus cycle to complete.

Look at the RESET and READY pins on the processor.  RESET show be low and READY should be
high.

-RST is tied to an RC circuit with a diode set up such that -RST is held low for ~47ms.  This comes directly from the Grossblatt book.

-AEN1 and -AEN2 are tied to GND.  RDY1 and RDY2 were left unconnected.  This also is from the circuit in the Grossblatt book.  After looking at it more, it does seem odd that those are left unconnected.  I tied both high through pull up resistors and pulled -ASYNC high as well.  The problem remained though READY is now high.  

You could also put a scope on the processor's clock input and make sure it looks
OK.  The old processors were pretty particular about voltage levels, i..e.
they weren't TTL compatible.  Hence the need for the clock generator chip.

Looking closer at the clock, it does seem like it is out of spec but I'm not sure why.  Attached is a capture from the scope on CLK.  Vtop (4.26V) > Vch of the 8088 (3.9V).  Vbas (856mV) seems too high for Vcl (0.6V).  I'm not sure what controls the bias on the clock voltage though to bring it down.

I've followed the circuit from the book and have done a few experiments (removing the cap in series with the xtal, tying RDY1 & 2 high, etc...) without much luck.  The one place where I haven't followed the circuit was the substituting out two 510ohm resistors tying the xtal pins to ground.  I've tried 470 and 560 ohm varieties as I don't have 510's on hand.  

Another note/oddity... I'm using a source voltage of 4.7v from a bench supply.  If I try to bump that up to 4.9 the 8284 seems to 'shutdown' in that the outputs go to noise.  Not sure what that's all about either.

The datasheet for the mbl 8088-2: http://www.datasheetarchive.com/Scans/Scans-000/Scans-0012900.html
The datasheet for the p8284a: http://www.datasheetarchive.com/pdf-datasheets/Datasheets-32/DSA-636967.html


This whole exercise is to try and understand more about circuit building itself so any thoughts or debugging help would be appreciated.

[Updated] Added attachment with DaveCAD schematic.

[pb]

I've been trying to debug this last night and tonight without much luck.  I must be missing something somewhere.  The 8184 datasheet says V-out-low should be .45V.  I believe this is for all outputs as the datasheet specially calls out V-out-high for CLK as being 4.0V while no such special callout for v-out-low.  Is that the correct interpretation?

If that is the case, I'm not sure where my 856mV for the CLK low value is coming from.  I've tried variations of the crystal circuitry (different crystal, etc...) with the same results.  I don't have another 8284 handy to see if it is that -- I was hoping to find a way to confirm that by investigation but am having no such luck.

Any thoughts?

Thanks!

[slburris]

I'd raise the voltage to 5 volts and work on why you get no clock then.
Running at 5 volts is the spec, so clearly something is not well with the
clock generator.  You shouldn't have to undervolt it to get things to work.
This issue has to be a major clue, although I can't suggest the failure
mechanism.  Certainly the .8 volts doesn't meet spec either.

Something else which doesn't make sense to me.  If the crystal is 14Mhz,
the oscillator output frequency should be that divided by 3, or 4.666Mhz.
Yet the Rigol trace shows 3.57Mhz.  The duty cycle should be 33%, but
the display shows 27.1%

Scott

[.o:0|O|0:o.]

I was also about to suggest bring VCC to 5V, 4.75V is the specified min. I can't see anything wrong with the configuration, but I noticed you are using XTALS. In the M8284A datasheet page 3/8 there is a paragraph suggesting the use of 10pF for best stability in certain cases. What type of board are you using? I would imagine that the deviation in the CLK frequency is the result of a poor X1, X2 input. You haven't mentioned trying the datasheet setup in Fig. 6, just that you used crystals. You could give that a go. Also, test the output on RESET which should be HI when RES' is pressed.

 .o:0|O|0:o.

[pb]

I'd raise the voltage to 5 volts and work on why you get no clock then.
Running at 5 volts is the spec, so clearly something is not well with the
clock generator.  You shouldn't have to undervolt it to get things to work.
This issue has to be a major clue, although I can't suggest the failure
mechanism.  Certainly the .8 volts doesn't meet spec either.

Something else which doesn't make sense to me.  If the crystal is 14Mhz,
the oscillator output frequency should be that divided by 3, or 4.666Mhz.
Yet the Rigol trace shows 3.57Mhz.  The duty cycle should be 33%, but
the display shows 27.1%

Scott

I'll look at this more tomorrow but I just retried running at 5V.  It seems to give me a clock out for a brief period of time after power-on at 5V and then flatline at ~.8V.  If I power-off/power-on again, I get that behavior consistently.

While it is running at 5.0V the result is even less pretty.  The .8V rises a smidge to 890mV but the frequency drops to 2.3 Mhz and the +duty drops to 17%.  If I drop the voltage even further to 3.8V I can get 33% duty cycle and Vbas is still around .8V.  Something seems fishy here.

Thanks for the pointers.

[pb]

I was also about to suggest bring VCC to 5V, 4.75V is the specified min. I can't see anything wrong with the configuration, but I noticed you are using XTALS. In the M8284A datasheet page 3/8 there is a paragraph suggesting the use of 10pF for best stability in certain cases. What type of board are you using? I would imagine that the deviation in the CLK frequency is the result of a poor X1, X2 input. You haven't mentioned trying the datasheet setup in Fig. 6, just that you used crystals. You could give that a go. Also, test the output on RESET which should be HI when RES' is pressed.

 .o:0|O|0:o.

I'm building this on a breadboard (elenco precision 9440 to be exact).  The Grossblatt book circuit suggests a 12-15pF cap in series with the crystal.  I've tried it with a 10pF or 15pF cap as well as without a cap with similar results.

I confess that I read that portion of the datasheet and am a little confused as to the intrinsic capacitance of breadboards and its effect on things.  In particular, are the frequencies involved here (14Mhz, 14/3Mhz, etc...) problematic.  Given the source material suggests breadboarding it I figured I was alright.  (you know what they say about assuming)

[Hewitson]

Can I ask where you sourced the 8088 from? Did you pull it out of some old PC, or are they still available?

I've pretty much mastered everything in MCU's that interests me, moving onto CPU's seems the next logical step. Ideally I'd be working with the 68k family, but I can't seem to get my head around the necessary logic to control them. The 8088 and 8086 seem considerably simpler.

[pb]

Can I ask where you sourced the 8088 from? Did you pull it out of some old PC, or are they still available?

I've pretty much mastered everything in MCU's that interests me, moving onto CPU's seems the next logical step. Ideally I'd be working with the 68k family, but I can't seem to get my head around the necessary logic to control them. The 8088 and 8086 seem considerably simpler.

I got the 8284 and the 8088 from digikey.  You can get 8088's from most places I looked.  The 8284's seem to be more rare.  I found two places that stocked them -- digikey and arcadecomponents.com

[Updated]  I don't know what I was thinking above.  I got the 8088 and 8284 from Jameco not digikey.

pb

[Hewitson]

I have found a site which you've probably already come across but I'll post it here for others if nothing else as I reckon its quite good..

http://www.helmpcb.com/Electronics/8088/8088.aspx

Looks like its very easy to set up a multiple processor system. Quite impressive for such an old chip.

[EEVblog]

I'm working on a (what I thought would be) simple 8088 based retro thing.  The Fujitsu 8088-2 that came from digikey doesn't have a notch or single dot index on it.  Instead it seems to have a groove/index along one side of the chip.  Based on the orientation and similarities to the ceramic package version in the datasheet, I believe that the indexed side marks pins 1-20.  Is that the correct interpretation?  Unless I'm missing something, I can't find explicit confirmation in the datasheet.

Yes, the groove/chamfer always indicates the pin1 side, on chips without the dot or other marking.
Exceptions may exist to this, but I can't recall ever having seen or heard of it.

Dave.

[pb]

I finally got around to getting another 8284a and the same circuit works now.  CLK, RESET, and READY look good now.

But... still no ALE from the 8088 after RESET.  I'll have to look at it more to figure out what is going on.  One step forward...

[pb]

Success!  I pulled everything from around the 8088 and rewired it trying to be more careful (and less aesthetic).  I'm beginning to think either my breadboarding skills are weak or my (nearly brand new) big breadboard is not so great for ... well .. breadboarding.

Thanks to all for the help debugging.