What if? Designing "Geneve 2020". Cool 3D views!

[+FarmerPotato]

35 minutes ago, RickyDean said:

this is a good idea, I'd like to plant this in my Rave motherboard with my Z80 card and my TI. Three computers in one! 

But how do they share the bus? 
 

[RickyDean]

@FarmerPotato, The Rave has the standard TIPEB bus with some extra slots, like a PC 16 bit bus that was to be used for future expansion. The motherboard has a interface card for the TI, and the Z80 plugs into a TI slot, only it uses just the power and ground pins, the rest is self contained. It can share floppy drives with the TI and Geneve, and was designed to be communicated with via terminal emulation, or an actual terminal. 

Edited November 8, 2023 by RickyDean
spelling

[+FarmerPotato]

2 hours ago, Jimhearne said:

I like using the 4 pin oscillator modules, if you fit an IC socket then really easy to change them for different frequencies for testing and overclocking etc. 

 

I like that.  Can you tell me part #s? I turned up DIP-14 ECS-100A-240 at Digi-Key.  ECS- 100-AX doesn't have a 24 or a 3. In fact I'm stuck finding a 3 MHz crystal for the 9902, and I don't want to divide the 24 because I will swap it! 
 

 

 

 

[+FarmerPotato]

Maybe dumb question:

 

In any datasheet, the crystal is shown connected to the CPU's two resonator pins. Under that is a schematic with a ~6 pF capacitor on each pin.

 

I understand that as a model of the normal parasitic capacitance. You don't actually put capacitors in there? The resonator doesn't require any more external components?

 

I've blindly added 6 pF disc caps in the past, no issues, and I tried my smallest 0805 22 pF last night. 

 

Just looking for confirmation that you don't actually put in those caps? 

[Jimhearne]

You need the capacitors unless you are using a 3 pin resonator, they have the caps built in.

A 2 pin resonator or a 2 pin crystal does need the caps.

The cap value depends on the crystal and the chip it's connected to, it should be specified on the datasheet but there is a fair tolerance , 15pf, 22 pf , sometime like that is typical.

 

Jim

 

 

[Jimhearne]

Actually, the 99105 datasheet does suggest 2 x 6pF , this is lower than a lot of chips, i see Stuart used 5pF on his project.

The TMS9995 datasheet suggests 15pF

 

Jim

 

 

 

 

Edited November 8, 2023 by Jimhearne

[Jimhearne]

36 minutes ago, FarmerPotato said:

I like that.  Can you tell me part #s? I turned up DIP-14 ECS-100A-240 at Digi-Key.  ECS- 100-AX doesn't have a 24 or a 3. In fact I'm stuck finding a 3 MHz crystal for the 9902, and I don't want to divide the 24 because I will swap it! 

 

 

I've a large amount of the crystal modules collected over the years, especially in the 4 to 80 Mhz range as they were used a lot on PC's in the 80's and 90's

Watch out for ones that have an enable pin but other than they they all seem to have the same pinout.

 

I've used just one 24Mhz crystal module feed to CLKIN and also my CPLD and divide it down by 6 or 8 (3 or 4 Mhz) in the CPLD for the 9901 and 9902 clock

You can use a 14 pin DIL socket and have the pinout able to take both the 8 pin size and 14 pin size modules with or without enable, turned pin sockets work best.  

 

Jim

 

 

Edited November 8, 2023 by Jimhearne
Remove rubbish Digikey search

[+FarmerPotato]

16 minutes ago, Jimhearne said:

Actually, the 99105 datasheet does suggest 2 x 6pF , this is lower than a lot of chips, i see Stuart used 5pF on his project.

The TMS9995 datasheet suggests 15pF

 

Jim

Alright, I'll put the caps back in and check again. 

My crystals and picoF caps are no-name brand. The oscillators I have on hand are 3.579 and 100 MHz..

I've got an onboard Pierce oscillator circuit, using  HC04 inverter. I'll play with that too. Also  have a '74 clock divider that I can switch in or out. (Solder link)

 

I like your idea of a PLD to divide the clock. 
 

I've got two loads on the CLKOUT pin:  the 9901 or  '74 divider,  the buffer to external bus. 
 

 

 

[Jimhearne]

4 minutes ago, FarmerPotato said:

Alright, I'll put the caps back in and check again. 

I've got an onboard Pierce oscillator circuit, using  HC04 inverter. I'll play with that too. Also I have a '74 clock divider that I can switch in or out. 
 

 

My crystals and picoF caps are no-name brand. The oscillators I have on hand are 3.579 and 100 MHz..

If you are using the HC04 oscillator then you just need to feed the output of that to XTAL1/CLKIN on the 99105 and nothing connected to XTAL2 , no need for the caps either.

 

You only need the caps if you are connecting the crystal straight across XTAL1 and XTAL2 on the 99105

 

Not sure which arrangement you have using or are your trying both ?

I thought from the picture of the PCB you were using the external HC04 oscillator.

 

Jim

 

[+FarmerPotato]

2 minutes ago, Jimhearne said:

Not sure which arrangement you have using or are your trying both ?

I thought from the picture of the PCB you were using the external HC04 oscillator.

 

Jim

 

I have both!  The Pierce oscillator was supposed to be for the 9902 (3 MHz crystal turns out to be scarce). The 99105 has the crystal underneath (short traces) and two 0805 caps. 
 

I could use the HC04 with 6 MHz, then divide CLKOUT. 

[Jimhearne]

Ah, i got you.

A search for  oscillator+module on Ebay turns up a lot of different modules, at least in the UK it does.

 

Jim

 

[+FarmerPotato]

App note from TI for Pierce oscillator using the cute little LVC1404 chip:

 

https://www.ti.com/lit/an/szza043/szza043.pdf

[Jimhearne]

Cute chip, not seen those before, certainly smaller than the oscillator modules.

 

Jim

 

[+FarmerPotato]

XTAL2 with 5 pF cap added to each leg of 12 MHz crystal. Oscillations do not get bigger than 370 mV Vpp and do not filter. They float around 2V.

 

The probes kill the signal when I try to measure the phase between XTAL1 and 2. 

 

red = RESET

yellow = XTAL2

 

This one happy event was after I changed probes to 10X attenuation (capacitance of probe interfering?)

Not getting this 95% of the time.

 

 

 

 

The noise is in my probe's ground loop. (I measured vcc vs gnd on the cpu, constant difference but both noisy on the scope). My gnd clips are on the power connector, next to 1 uF and 10 uF electrolytic caps.

 

 

 

 

[+FarmerPotato]

 

 

 

Reset is an RC circuit into two Schmitt trigger inverters (LS14). Here's 3.5 ms between power-on and reset high:

 

[Jimhearne]

A scope probe on X1 will kill the oscillator as you've found, X10 is normally ok , you should see the oscillation on at least 1 of the 2 crystal pins though it will be a low level, don't expect it to be 5V pk-pk.

I have my 990/10A setup here, i will scope the clock pins on that.

 

 

[Jimhearne]

Sorry, i'm being silly, the 10A doesn't have the crystal on the processor.

 

 

[Jimhearne]

I know there are different types of crystals according to the way they are cut i think, but I've not had a problem before just using what i have.

It maybe that the ones you have just aren't compatible with the 99105 oscillator circuit.

 

 

 

[+FarmerPotato]

1 hour ago, Jimhearne said:

I know there are different types of crystals according to the way they are cut i think, but I've not had a problem before just using what i have.

It maybe that the ones you have just aren't compatible with the 99105 oscillator circuit.

 

 

 

I'll keep fiddling with it, and test the Pierce oscillator HC04.

Other crystals from this batch were ok on breadboard or in my small PCB. (as was this 99105 chip.)

 

Wondering if temp can ultimately disable  a crystal? To minimize the traces, I put it (HC49 half tall)  within  the CPU's socket frame. (Should have widened the traces too.) I bet it gets hot under the CPU.

 

A benefit: all my tools finally set up in the new room. Yesterday I was using an ancient Micronta logic probe!

 

[Shift838]

19 minutes ago, FarmerPotato said:

I'll keep fiddling with it, and test the Pierce oscillator HC04.

Other crystals from this batch were ok on breadboard or in my small PCB. (as was this 99105 chip.)

 

Wondering if temp can ultimately disable  a crystal? To minimize the traces, I put it (HC49 half tall)  within  the CPU's socket frame. (Should have widened the traces too.) I bet it gets hot under the CPU.

 

A benefit: all my tools finally set up in the new room. Yesterday I was using an ancient Micronta logic probe!

 

I can confirm that too much heat can kill a crystal.  I've done it!

[Jimhearne]

Heat from soldering can kill a crystal but heat from the CPU wouldn't be a problem unless the whole thing was marginal in the first place.

I notice there is a Via on the board near one of the crystal pins, if the can of the crystal is touching that (assuming vias aren't tented) then it might cause issues.

Crystals should ideally be spaced up from the pcb just a touch to stop the possibility of the can shorting out to any tracks running below them. 

A bit of cardboard (0805 component tape) under them is plenty.

I don't think it's any of those things but just thoughts.

 

Jim

 

[+FarmerPotato]

1 hour ago, Jimhearne said:

I notice there is a Via on the board near one of the crystal pins, if the can of the crystal is touching that (assuming vias aren't tented) then it might cause issues.

Yeah, I checked for shorts with that via. I'll check again. No tented or buried vias. I'll replace the crystal again and inspect underneath. 

Thanks for looking at it closely!

 

I wanted ground fill (back layer) under the 99105.  I did some questionable signal routing to keep it contiguous. (Also I landed the CPU at the edge of the board, I won't do  that next time.) 

Folks tell me "just fill the planes" and I know from book theory that the ground plane fragmentation (slots) matters at 20 MHz.

 

I'll try to simplify ground returns next time. I admire TI's grid of chips with regular Vcc and GND rails.  That was in thru hole days. 

[+FarmerPotato]

12 pF and 12 pF on pins: This oscillation starts consistently but lasts for about 3 s.

Nice phase separation. Amplitude is overshooting 5 Vpp. (dashed cursor lines on 0 and 5)

 

I lifted the 0805 pad trace (4th time soldering to it?) Soldered wire leg to crystal. Perhaps extra pF.

 

 

 

17 pF and 12 pF on pins: This oscillation starts consistently but still dies.  The amplitudes relate to different pin capacitance.

 

 

Must stop for now.

 

[Jimhearne]

Very strange it starts ok and then stops.

What do the 2 pins do when it stops.

I'm sure you've checked but it's not loosing the 5V supply after 3 seconds ?

Both ground pins connected on the 99105 ?

 

[+FarmerPotato]

38 minutes ago, Jimhearne said:

Very strange it starts ok and then stops.

What do the 2 pins do when it stops.

I'm sure you've checked but it's not loosing the 5V supply after 3 seconds ?

Both ground pins connected on the 99105 ?

 

When it stops, both pins float at 1.6.  I didn't notice a big deviation.  Yesterday there was the steady state with 370 mVpp oscillation around 1.88V.