My 6502 Roadmap - Comments, Advice and Suggestions Requested

[floobydust]

Shawn Odekirk wrote:

I got floobydust's SBC1 C02Monitor V1.04 monitor running. Thanks floobydust! There's a ton I can learn from your code.

Glad to hear that you have the code running on your hardware. I think we've all learned a lot from each other's code over the decades. Hopefully the comments in my source is useful... there's also a CMOS only version of Enhanced Basic which will run with the V1.04 Monitor.

I've also moved off the 65(C)51 UARTs... my latest C02 Pocket-SBC uses a NXP/Philips SCC2691 UART. Granted, it's an older chip and limited to ~6MHz CPU clock, but the Monitor and BIOS for this SBC is separate and has several updated routines and functions, including Xmodem-CRC upload. At some point, I might create a new BIOS for the SBC1 so it can use the new Monitor version.

In any case, have fun and keep posting updates!

[Shawn Odekirk]

My order of 2 PLCC-44 wire wrap sockets arrived the other day, opening up the possibility of trying the NXP 28L92. Before placing an order I'd just like to ask if I am forgetting anything.

• NXP 28L92
  3.6864 MHz can oscillator
  MAX 238
  4 1uF tantalum capacitors for the MAX 238

Thanks in advance!
Shawn

[floobydust]

Shawn Odekirk wrote:

My order of 2 PLCC-44 wire wrap sockets arrived the other day, opening up the possibility of trying the NXP 28L92. Before placing an order I'd just like to ask if I am forgetting anything.

• NXP 28L92
  3.6864 MHz can oscillator
  MAX 238
  4 1uF tantalum capacitors for the MAX 238

Thanks in advance!
Shawn

Also, a pair of 24pF caps for the crystal and note that the NXP UARTs have a RESET that is active going positive, not negative, so you'll need an inverter from the standard reset signal for the 6502.

BDD likely has the most experience with the 28L92, but there are others as well. I did a full support BIOS for the earlier single channel SCC2691 which supports the timer as a jiffy clock, received break and transmit/receive all via interrupt. I used the SCC2691 on my recent C02 Pocket SBC. As you're running on a 65C02, you can likely modify the BIOS to run one channel of the 28L92 as a console. Also, I use a FTDI USB-UART adapter instead of a normal RS-232 level convertor. It interfaces directly to the UART and gives you a USB port for any computer... i.e., OSX, Linux or Windoze.

[BigDumbDinosaur]

floobydust wrote:

Also, a pair of 24pF caps for the crystal...

Say what? He has a can oscillator in his shopping list. All that needs is the usual bypass capacitor, a 0.1 µF XR7 MLCC.

[BigDumbDinosaur]

Shawn Odekirk wrote:

My order of 2 PLCC-44 wire wrap sockets arrived the other day, opening up the possibility of trying the NXP 28L92. Before placing an order I'd just like to ask if I am forgetting anything.

• NXP 28L92
  3.6864 MHz can oscillator
  MAX 238
  4 1uF tantalum capacitors for the MAX 238

Thanks in advance!
Shawn

You need five capacitors for the MAX238, four for the charge pump circuitry and one to bypass the device. All five capacitors should be as physically close to the MAX238 as possible.

BTW, I've done some testing and have determined that it is possible to use ceramic capacitors for this application. I used Kemet's part number C315C105K3R5TA, which is a 100 mil lead-spacing part that is approximately the same physical size as the tantalum part.

[Shawn Odekirk]

floobydust wrote:

and note that the NXP UARTs have a RESET that is active going positive, not negative, so you'll need an inverter from the standard reset signal for the 6502.

I think have have some spare pins on a G22V10 that I can use for that.

floobydust wrote:

Also, I use a FTDI USB-UART adapter instead of a normal RS-232 level convertor. It interfaces directly to the UART and gives you a USB port for any computer... i.e., OSX, Linux or Windoze.

I have an FTDI adapter that I'm using currently, but I'm thinking ahead for when I have a PCB made. It seems like a DB-9 serial port would be more appropriate for a PCB than the raw UART lines. Correct me if I'm wrong.

BigDumbDinosaur wrote:

Say what? He has a can oscillator in his shopping list. All that needs is the usual bypass capacitor, a 0.1 µF XR7 MLCC.

I'm not sure what you are referring to. I plan to have 0.1 µF bypass capacitors for all of the ICs. Does the can oscillator need a bypass capacitor from Vcc to Gnd as well? No problem if it does. I can easily add one.

BigDumbDinosaur wrote:

You need five capacitors for the MAX238, four for the charge pump circuitry and one to bypass the device.

I was going to go with a 1 µF electrolytic capacitor as the bypass for the MAX 238, per your "INTERFACING THE NXP SC28L92 DUART" document. Sorry I forgot to include that in my shopping list.

BigDumbDinosaur wrote:

BTW, I've done some testing and have determined that it is possible to use ceramic capacitors for this application. I used Kemet's part number C315C105K3R5TA, which is a 100 mil lead-spacing part that is approximately the same physical size as the tantalum part.

I'm going to show my ignorance here. Is there a benefit to using ceramic capacitors rather than tantalum? My decision to use tantalum capacitors for the MAX 238 charge pumps also came from reading your interfacing document.
I'm sorry, but I just don't know the characteristics of the various capacitor types in order to know when one would be more appropriate to use than another. Can the 0.1 µF bypass capacitors for all of the ICs be tantalum or should they be ceramic?

Thanks for the replies!

[Chromatix]

Ceramic capacitors have the lowest ESR (equivalent series resistance) at high frequencies. They're also very cheap and robust. This makes them ideal for bypassing high-frequency currents in power supplies. I would use them just about everywhere that they make economic sense.

Tantalum capacitors also have quite low ESR, and tend to be available in higher capacities than ceramics. They do however tend to burst if subjected to a sharp inrush current, so you have to pay attention to where you use them (ie. not directly connected to an external power input). They're safe to use on the output side of a switch-mode power supply or anything with a current limiter, including in the voltage boost circuits of an interface chip.

Electrolytic capacitors have relatively high ESR, especially in the higher capacities, and are almost always polarity sensitive. But they also cover the highest capacity ranges. Use them in combination with low-ESR capacitors for bulk power smoothing, and little else. You could reasonably parallel up a 1000µF bulk electrolytic, a 100µF low-ESR electrolytic, and a 10µF ceramic at your power input, and then use 100nF ceramics to locally bypass individual ICs.

[GARTHWILSON]

There's a good 49-minute talk on capacitors from a Kemet engineer at https://www.youtube.com/watch?v=ZAbOHFYRFGg .

I expect the best to use for the line-driver charge pumps would be ceramic. The data sheet probably has recommendations for type, maybe even for exact part numbers.

[Chromatix]

In general, I would recommend keeping a decent stock of 100nF and 10µF ceramic capacitors, as well as a few of whichever picofarad-level values keep cropping up in crystal oscillator and similar stuff.

[floobydust]

BigDumbDinosaur wrote:

floobydust wrote:

Also, a pair of 24pF caps for the crystal...

Say what? He has a can oscillator in his shopping list. All that needs is the usual bypass capacitor, a 0.1 µF XR7 MLCC.

Ahhhh.... Oops. Somehow I just read crystal into it... don't ask me why, I don't know either

[BigDumbDinosaur]

Shawn Odekirk wrote:

floobydust wrote:

and note that the NXP UARTs have a RESET that is active going positive, not negative, so you'll need an inverter from the standard reset signal for the 6502.

I think have have some spare pins on a G22V10 that I can use for that.

That's kind of wasteful of the GAL's limited resources, don't you think?

Quote:

I'm not sure what you are referring to. I plan to have 0.1 µF bypass capacitors for all of the ICs. Does the can oscillator need a bypass capacitor from Vcc to Gnd as well? No problem if it does. I can easily add one.

All active devices should be bypassed. That includes oscillators.

Speaking of oscillators, be sure the one you get produces CMOS output levels, not TTL.

Quote:

BigDumbDinosaur wrote:

You need five capacitors for the MAX238, four for the charge pump circuitry and one to bypass the device.

I was going to go with a 1 µF electrolytic capacitor as the bypass for the MAX 238, per your "INTERFACING THE NXP SC28L92 DUART" document. Sorry I forgot to include that in my shopping list.

The MAX238's bypass capacitor should be a tantalum or ceramic. I also put a 100 µF electrolytic in parallel with the ceramic/tantalum bypass capacitor. The tantalum/ceramic takes care of bypassing the high frequency noise that the MAX238 kicks back into the power supply, while the electrolytic provides some voltage "cushion" as the MAX provides current to drive its outputs to the serial link.

Quote:

Quote:

BTW, I've done some testing and have determined that it is possible to use ceramic capacitors for this application. I used Kemet's part number C315C105K3R5TA, which is a 100 mil lead-spacing part that is approximately the same physical size as the tantalum part.

I'm going to show my ignorance here. Is there a benefit to using ceramic capacitors rather than tantalum? My decision to use tantalum capacitors for the MAX 238 charge pumps also came from reading your interfacing document.

At the time when I wrote that I had not had an opportunity to compare operation with ceramic capacitors against operation with tantalums. As previously noted, ceramics in that capacitance range are less expensive than tantalums and have the lowest ESR of the commonly-used capacitor types. Also, ceramics are not polarized like tantalums, so there is no danger of you accidentally putting one into the circuit backwards and having it suddenly fail.

Quote:

I'm sorry, but I just don't know the characteristics of the various capacitor types in order to know when one would be more appropriate to use than another. Can the 0.1 µF bypass capacitors for all of the ICs be tantalum or should they be ceramic?

Bypass capacitors should be MLCC XR7, 0.1 µF, 50 volts. When you design your board a key objective is to plan the layout so the connection between the bypass capacitor and the Vcc pin of the device being bypassed is as physically short as possible. That also implies that the capacitor's leads be ket as physically short as possible.

[Shawn Odekirk]

Thanks everyone for your comments! I've placed my order and will post an update after my parts arrive I've made some more progress.

Shawn

[Shawn Odekirk]

I have replaced my 1 MHz oscillator with a 10 MHz oscillator and everything seems to be working fine, which is a little surprising since the RAM and ROM are 150 ns parts and I have not implemented wait-states. I have written a small RAM test program and it seems to work correctly.
What kind of stress testing can I do to see how close to the edge I am?

[GARTHWILSON]

Quote:

I have replaced my 1 MHz oscillator with a 10 MHz oscillator and everything seems to be working fine, which is a little surprising since the RAM and ROM are 150 ns parts and I have not implemented wait-states. I have written a small RAM test program and it seems to work correctly.
What kind of stress testing can I do to see how close to the edge I am?

See this post and the ones following it, and
also this post and the ones following it.

[cbscpe]

Shawn Odekirk wrote:

I have replaced my 1 MHz oscillator with a 10 MHz oscillator and everything seems to be working fine, which is a little surprising since the RAM and ROM are 150 ns parts and I have not implemented wait-states. I have written a small RAM test program and it seems to work correctly.
What kind of stress testing can I do to see how close to the edge I am?

Use a strong aka hot hairdryer and heat your setup or if you have an oven that supports low temperatures like 60 celsius put the setup in the oven