The transformer's magnetic field probably won't cause any trouble but the EMI it will emit may.

The trouble with your plan to use a linear power supply is the lack of crowbar capability in case something goes to ground. The 7805 regulator will current-limit if subjected to a direct short but may not respond to a low resistance fault (unless >1.1 amps is drawn). I'd think about this some more if I were you. You can get inexpensive switching power supplies in bare board form that will provide what you need, along with crowbar capability.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

Having the transformer close won't be any problem. The magnetic field is pretty well contained in a closed loop in its core. Power transformers are routinely used in audio equipment that has to have a very low noise floor. Where you have to be more careful in design and construction is keeping digital noise out of analog circuits.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

As for short circuits, i doubt that shortcuts will manifest, and if they do i plan to use fast fuses(fast, but they are not that fast, i tried to blow one once, it takes a while with a current that is not much bigger that the fuse declared current). Also, i could try to limit the current that goes to the 2N3055 base, so if an overload or shortcut occurs it would give a constant maximum current (2A).

The purpose of a crowbar circuit is not to protect the power supply against a short circuit (That's what current limiting does.) A crowbar is used to protect the powered system from a failure of the power supply. Should the output voltage rise too high (say 6.25V ina 5V system), the crowbar circuit puts a massive short across the output of the power supply, blowing fuses (and probably the rest of the power supply as well, but if the crowbar goes, it had failed anyway!). The idea is to kill the power supply before the output voltage rises high enough to damage the circuit it is powering.

A 7805 regulator's internal current limiting will protect it from a short circuit or other high-current situation, providing the device is mounted on a big enough heatsink - check the datasheet. Adding an outboard pass transistor (eg 2N3055) will effectively disable this current limiting (it will still work, but at an undefined current of tens of amps, the exact value dependant on the gain of the transistor).

I've used 7085's to power my homebrew devices for 30 years now and never had one fail from overloading. The only one that has failed died from a massive spike on the incomming mains line - adding a mains spike suppressor prevented this happening again.

If one 7805 isn't sufficient, you can always use several, each supplying just a part of the circuit - don't connect the outputs together or you will have problems. The Acorn Atom used two 7805s in this manner and I understand that S100 cards used to have a regulator on each card.

_________________
Shift to the left,
Shift to the right,
Mask in, Mask Out,
BYTE! BYTE! BYTE!

As for voltage rising over 5V, i used zener diodes in my last power supply, i also tested it, and it works, the idea is that the fuse would blow before the zener diode(but I didn't prove it)...

In my sbc i am using the 28C256 eeprom, and i think this model has the capability to selectively rewrite each address location, so i thought that it would be nice if it would be possible to reprogram my sbc rom via serial, without having to remove the eeprom and program it via programmer. Has anyone tried to do something like this?

Yes. I did that on my SBC-2.

You have to keep ALL of your actual code in RAM during the write process. That included device drivers and data tables. You will be unable to read anything from EEPROM while a write sequence in underway.

Other than that, its straight forward.

Daryl

I am glad this can work well, i will implement this in my sbc, but i will also put a write protect switch.

What is the easiest way to get -5V for VEE(for the dac)?

Since you've not specfied the current level required, regulation accuracy, and noise levels, I'll recommend to you the 7660 series of switched capacitor inverter circuits. They are not well regulated, and can generally provide only between 20 and 50 mA depending on the size and quality of the capacitors you use. The Microchip TC7660 is a candidate, but Analog Devices (ADM660), Maxim (MAX660), the TI LM2660 are also candidates.

Depending on your application for the bipolar DAC, the negative supply does not have to all that accurate or capable of providing a lot of current. These devices are switching, inductor-less regulators, so they will be too noisy for most percision analog work. They can however be used to invert an input ranging from 1.5V to 10V. Thus, if you can stand the losses and the additional cost, then you could follow the inverter with a negative voltage linear regulator (such as an LM7905) to clean up the power somewhat.

National Semi also makes a very easy to use line of switching regulators that you can configure to generate a negative voltage: look at their Simple Switcher line. Again, you may require a linear post-regulator to clean up the generated noise for any precision analog work.

Finally, depending on ypur application, you could also take advantage of the -6V output of most switched capacitor RS-232 drivers. You will not be able to pull much current from these devices. In my experience, Analog Devices has the best switches (fastest switching times with the lowest channel resistance) in these types of devices, and so will provide the cleanest output, and the most current.

_________________
Michael A.

Interestingly since MichaelM mentioned the RS-232 line driver ICs with onboard switched-capacitor voltage boosters, I recently read in an article that that's exactly where the switched-capacitor power supply ICs originated. Manufacturers found out that designers were incorporating these into circuits not for the line-driver capability, but for the switched-capacitor voltage convertor.

The 7660 line has a lot of more-modern relatives that are much better. I have data sheets for a load of them downloaded. Even among the best of them, you have to choose between different trade-offs, like if you want built-in regulation like the TI LT1054 which I have a module made with here that has 0.67Ω output impedance so an 87mA load pulled it down from -4.95V only to -4.89V (!) (using a 9V battery for input), or the National Semiconductor LM2662 which has an output impedance of 3.5Ω but can achieve 86% efficiency @ 200mA with 5.5V input (I have a module built up with that too), or the LTC1144 that works to 18V, etc.. These are switched-capacitor voltage converters, and good performance depends partly on good layout and good capacitors that have very low impedance at frequencies starting in the tens of kHz. Ceramic are good, and OS-CON (Organic SemiCONductor, a type of solid-electrolyte capacitor made by Sanyo and a couple of other companies) are even much better than tantalum.

+1!


Using good capacitors and layout, I've found that they can be pretty good. I tend to add large ferrite beads to switching supplies' inputs and outputs, but even without them, I'm getting around 5mV peak-to-peak of switching noise on the outputs.

Another thing I've done is make a square-wave oscillator with a Schmitt-trigger inverter and use that to drive more of them, paralleled for more drive capability (74AC14 is much stronger than 74HC14) and then run their output into a voltage multiplier made of capacitors and Schottky diodes, and then follow that with an analog regulator if necessary. I have a module here I made like that which starts with +5V and puts out a few watts at -11.5V to get regulated down to -9V with a linear regulator.

The Power Trends integrated switching regulators are nice too. It's just too bad that both they and National Semiconductor have been acquired by TI.

Another module I was thinking of offering for sale for hobbyists is one that takes in 7.5-18V (so you could even run your project from automotive power) and put out +5 and ±9V. The trouble I run into is that without enough sales volume to justify the tool-up cost for automated assembly, the labor makes it economically prohibitive, and I can't very well offer it as a kit either since switching power supplies (which it would partly use) do not lend themselves to hand assembly without special tools. I might still find a way.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

Thanks for the elaborate replies.
I bought the DAC0806LCN from Unicorn electronics because i had to fill the 25$ limit, when i was ordering the chroma modulator for the crtc.
This dac requires -13mA max for VEE, first i have to see if they sell the ic that you suggested in the local electronic stores at all, if not... ebay...

I've used the DAC0808 a lot which is the same thing but more accurate, and I use typically -9V but it's not critical at all. I've used it as a multiplying DAC too, which means similar to a digital pot.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

I was planning to build an audio module, with this dac, and with an adc. The module should have 512k of ram, acting like a buffer, and also a digital i/o (so it can also be used as a logical analyzer). Also there will be a address counter with a clock source (like a local dma), so that the cpu doesn't have to copy the bytes from the ram banks to the dac/adc. I was thinking about using one 9536 cpld, which should handle the data transfer and addressing....

My experience has been that tantalum capacitors are well suited to this sort of application. They have low losses relative to the available capacitance.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!