Trivial circuits for sbc

[Dajgoro]

Problem solved! The serial cable that i have been using was a mess, i rearranged the wires in the connector, and now it works...

[Dajgoro]

Once i found a simple and clean schematic for implementing the halt and single step debugger circuit. I placed two switches on the chassis, halt and step, so i am looking for a simple circuit that would do the trick...

[BigDumbDinosaur]

GARTHWILSON wrote:

...BDD: And about the capacitors, look into OS-CONs. You'll be pleasantly surprised (like I was when you told us about 74ABT) that OS-CONs are even a lot better than tantalum. See http://www.edc.sanyo.com/pdf/e028.pdf

I have looked at them. Assuming one is using the recommended capacitor sizes with the MAX232, 238, etc. (typically 1 µf), I don't see where the OS-CONs offer a lot over the equivalent tantalums. The OS-CON ESR vs. temperature curve is definitely better than the tants and high ripple frequency characteristics are substantially better. However, most of these characteristics are not a critical factor in the MAX's charge pump circuit performance.

On the other hand, I see where OS-CONs would be a real asset in switch mode power sources. I suspect they were targeted to wall-wart switch mode devices, where small size as a function of capacitance and excellent ESR, ESL and temp characteristics would be very advantageous.

[GARTHWILSON]

I recently heard also that tantalum-capactitor manufacturers tried a few years ago to put a ring in the nose of the buyers who needed the characteristics which tantalum used to kind of have the corner on, and force market prices sharply upward, but that the industry responded by developing multilayer ceramic chip capacitors (MLCCs) in much higher capacitances to take their place, pulling the rug out from under the tantalum-capacitor manufacturers. So we now regularly use MLCCs in 4.7uF/25V 1206, and you can get even more than that in that case size. They're not as cheap as aluminum electrolytics yet, but their MTBF is way longer, which the military is interested in. For the little charge pumps, those might be the way to go if you're laying out an SMT board. I have not compared them to OS-CONs yet for that application; but yes, we use the OS-CONs for switching supplies where we still need the greater capacitances.

[BigDumbDinosaur]

GARTHWILSON wrote:

I recently heard also that tantalum-capactitor manufacturers tried a few years ago to put a ring in the nose of the buyers who needed the characteristics which tantalum used to kind of have the corner on, and force market prices sharply upward, but that the industry responded by developing multilayer ceramic chip capacitors (MLCCs) in much higher capacitances to take their place, pulling the rug out from under the tantalum-capacitor manufacturers.

I heard something like that as well but don't know if it's true. I do recall a momentary spike in tantalum capacitor prices sometime ago and then suddenly MLCCs started appearing. It could have just been a coincidence though. At least we have more choices and if space isn't at a premium I use MLCCs.

Quote:

So we now regularly use MLCCs in 4.7uF/25V 1206, and you can get even more than that in that case size. They're not as cheap as aluminum electrolytics yet, but their MTBF is way longer, which the military is interested in. For the little charge pumps, those might be the way to go if you're laying out an SMT board. I have not compared them to OS-CONs yet for that application; but yes, we use the OS-CONs for switching supplies where we still need the greater capacitances.

On the POC boards, I'm using through-hole capacitors, so the tants have a size advantage over MLCCs of the same capacitance. I can't work with the really small SMT stuff because of deteriorating vision...hell, deteriorating everything!

[Dajgoro]

Its maybe a silly question, but why the nmos 6502 has dynamic registers(with the minimum frequency limit)?

[BigEd]

Dajgoro wrote:

Its maybe a silly question, but why the nmos 6502 has dynamic registers(with the minimum frequency limit)?

Because their whole business idea was to hit a very low cost point, and a single transistor to hold a bit of state is the smallest area possible (it's DRAM) - cost of silicon is an exponential function of area, so their area target was very strict.

It's one of the reasons CMOS was a bit slower in adoption than it might have been: there's a power dissipation advantage, but the area and therefore cost penalty is high. If the manufacturer isn't able to pass on the cost, then they have a reduced profit margin and a reduced incentive. (The RCA COSMAC is an outlier, but if it was selling into a space/military market, or indeed into a market where low power is worth paying for, it's worthwhile.)

Cheers
Ed

[BigDumbDinosaur]

BigEd wrote:

It's one of the reasons CMOS was a bit slower in adoption than it might have been: there's a power dissipation advantage, but the area and therefore cost penalty is high. If the manufacturer isn't able to pass on the cost, then they have a reduced profit margin and a reduced incentive. (The RCA COSMAC is an outlier, but if it was selling into a space/military market, or indeed into a market where low power is worth paying for, it's worthwhile.)

Cheers
Ed

There was also the matter of performance. The early CMOS devices were substantially slower than their NMOS counterparts.

[Dajgoro]

When writing data into eeprom-s the write cycle lasts a few ms. But do i have to keep the data address and write signals during that few ms, or can i write into it like in a static ram, and then just do nops until it is done? I took a 29F002N eeprom (actually the bios chip) from an old motherboard and i am planning to use it in my sbc for storing data/programs.

[leeeeee]

You write a write command, write the data bytes and then poll the busy bit until the write completes.

Lee.

[Dajgoro]

Quote:

To write a command or command sequence (which
includes programming data to the device and erasing
sectors of memory), the system must drive WE# and
CE# to VIL, and OE# to VIH.

There is also no busy pin.
So i just write like it was ram, and do a few nops?

[leeeeee]

The busy bit, b6 at any address, will toggle between 0 and 1 on subsequent reads until a command is complete.

Lee.

[Dajgoro]

I am trying to get a simple way of programing the eeprom, i don't mind if it is slow...
So i will put it this way: Will the data be stored in the eeprom if i write pretending it was a static ram, and then waiting that few ms that are required for the data to be permanently programed into the floating gates?

[leeeeee]

Quote:

Will the data be stored in the eeprom if i write pretending it was a static ram

No.

You start with by writing a write command.

You then write up to a page of data writing each byte within the maximum time from the last byte, usually 100us.

You then either wait for the maximum write time or poll the busy bit. Polling is faster and safer.

You cannot read from the device until the write command is completed.

Lee.

[Dajgoro]

I got 0.1mm copper enameled wire from ebay, and it looks kinda thin. It has about 3ohm/meter, and now i am wondering if it would work well if i used it instead of the ribbon cable that i usually use for building sbc-s.