EEVblog #238 - Power Supply Design Part 7.

[firewalker]

Alexander.

[hans]

I think it may be possible to reduce the component count by removing the pull-up resistors on the input switches. I have used the MCP23S17 before for reading switches (for a joystick) and used the internal pullups. The datasheet also mentions and documents a register for it (register GPPU)

The uCurrent seems like very sensitive to me. 500 gain introduces a ton of noise. Let's not forget that the output noise of the opamp is 1.5uVp-p (RTI). If you amplify it by 500, that's about 0.75mVpp noise. The ADC samples 0.5mV steps, so it will require a bit more oversampling to get a completely 'silent' readout. Nevertheless, more resolution = better.

[thilo]

It turned from a nice little power supply in a feature creeped monster 

[firewalker]

Dave would it be nice to use opamps with supply voltage at least to 22~30 volts in order to enable people to modify it easily for higher voltage?

Alexander.

[ejeffrey]

Dave, what is the reason to use soft SPI instead of sharing the pins with the ISP?  I would have assumed that as long as the chip-select pins were off during programming the other devices on the bus wouldn't interfere with the programming.

Also, if you were starting over would you reconsider using a capacitive switcher in to generate a negative rail?  It seems like even a 10 milliamp negative supply would have made a number of issues easier, but presumably that would come with its own issues.

[mightyzen]

Dave, I think you misread the TLC272 data sheet. Texas did not mix up A, B and C grades. The TLC272CD is not the C-grade, it is the grade-less. C = 0-70C, D = SO-package. The  TLC272ACD is the better A-grade, the  TLC272BCD is the even better B-grade. There is no C-grade.

Why don't you add a simple negative voltage generator (eg TC7660) to your design? Then you could use the cheapest standard op amps, like the lm258 or ka258.

[jpelczar]

Dave, what is the reason to use soft SPI instead of sharing the pins with the ISP?  I would have assumed that as long as the chip-select pins were off during programming the other devices on the bus wouldn't interfere with the programming.

Or one could use 4053 like here: http://www.btc.pl/pdf/zl12avr.pdf

[fmaimon]

You can´t and don´t need to leave that 100nF capacitor on the reset pin. Just a single 10k resistor is enough. The AVR reset have a internal filter to discard noise and a capacitor will disturb the signals going to/from reset if you want to use debugWire in the m168.

[McMonster]

Dave, what is the reason to use soft SPI instead of sharing the pins with the ISP?  I would have assumed that as long as the chip-select pins were off during programming the other devices on the bus wouldn't interfere with the programming.

Also, if you were starting over would you reconsider using a capacitive switcher in to generate a negative rail?  It seems like even a 10 milliamp negative supply would have made a number of issues easier, but presumably that would come with its own issues.

I don't think it'll matter now, Dave mentioned in other thread the project already went into production. But, should it matter for any possible future revisions it's much simpler. There's an AVR042 application note about hardware considerations for AVR chips with states that putting series resistors on SPI lines is all you need for proper ISP functioning even if there are other devices trying to drive those lines. So in my opinion it's worth considering if there'll be any new board version, hardware interface means simpler code.

And great work with expansion connectors, people can't really complain on the lack of modding possibilities. I just hope nobody will mention arduino shields now, there's simply no point

[Bored@Work]

There's an AVR042 application note about hardware considerations for AVR chips

I previously pointed Dave to it, he didn't seem to be much interested. I was also surprised to hear his reasoning for changing the LED on the reset pin "after reading the datasheet". While this was in fact something pointed out be me in the forum https://www.eevblog.com/forum/index.php?topic=6163.msg80321#msg80321

[firewalker]

I believe he did it in order to have the SPI available for any kind of Arduino control modules that need it. 

Alexander.

Dave, what is the reason to use soft SPI instead of sharing the pins with the ISP?  I would have assumed that as long as the chip-select pins were off during programming the other devices on the bus wouldn't interfere with the programming.

Also, if you were starting over would you reconsider using a capacitive switcher in to generate a negative rail?  It seems like even a 10 milliamp negative supply would have made a number of issues easier, but presumably that would come with its own issues.

[mikeselectricstuff]

Don't know where you got the pinout for the UART ins on your header but most of the  ones I've seen use the pinout of FTDI's TTL-232 cable, which doesn't match what you have there.
 

[amspire]

And great work with expansion connectors, people can't really complain on the lack of modding possibilities. I just hope nobody will mention arduino shields now, there's simply no point

Sorry to dash your hopes, but I still think that to eliminate the chance to use a Ethernet interface is a big mistake. And the only most practical way of adding Ethernet is to use the $20 (delivered) shield.

Any computer interface has to be isolated, and the serial based USB is just ugly compared to USB.

Ethernet means you don't have to have a computer on the same bench, you get 1500V isolation, you do not need a program to talk to the power supply since you can use a browser, you get the SD socket, so you can have unlimited storage space for data logging, you never have to sort out some piece of software that has to work out what virtual serial port the USB is using, since it changes for each different USB port, if you have a wireless router on your network, you get wireless access from notebooks, ipads, phones, etc, you can have banks of power supplies that can all talk to one computer.

You get all of this merely by freeing up a few more pins.

In case anyone thinks accessing the power supply via a browser means some heavy duty stuff - that is wrong. A basic web server is one of the simplest programs you can write. You listen for a GET command string, and send back your data along with some stored HTML strings. To get started, that is all that is needed.

Now that Ethernet seems to have been ruled out, someone will have to design the isolated RS232 or USB ports, and they will either have to use a terminal program (missing in Windows 7), or actually write a custom interface program. All that extra work for the sake of 2 pins on the Arduino.

Richard

[firewalker]

In what way the ethernet shield is ruled out?

Alexander.

[jerry507]

There's an AVR042 application note about hardware considerations for AVR chips

I previously pointed Dave to it, he didn't seem to be much interested. I was also surprised to hear his reasoning for changing the LED on the reset pin "after reading the datasheet". While this was in fact something pointed out be me in the forum https://www.eevblog.com/forum/index.php?topic=6163.msg80321#msg80321

Really that hard up for the recognition?  The statement certainly didn't imply that he found it after that, just that it was not something obvious and was "buried in the datasheet."

[Bored@Work]

Really that hard up for the recognition? 

I am just surprised. Dave did a number of corrections as suggested by forum members. I would have expected a nod to the forum. Of course I do understand he needs to keep up a certain image in his videos.

[amspire]

In what way the ethernet shield is ruled out?

Alexander.

Looking again, and you may be right. If the pin labelled SS is unused, then the Ethernet part of the Ethernet shield at least could be used. For some reason I thought that pin had been assigned to PS hardware, but I think I was wrong.

The SD memory enable pin (PD3) has been allocated for the rotary encoder. I guess if another spare pin can be found, you could edit the pin allocation in the library and regain access to the SD slot.

Does anyone know where the link to the new schematic PDF is? I tried the http://www.eevblog.com/uploads/ folder but only RevA is there right now.

[Psi]

Really that hard up for the recognition? 

I am just surprised. Dave did a number of corrections as suggested by forum members. I would have expected a nod to the forum. Of course I do understand he needs to keep up a certain image in his videos.

It's not scripted, he can't be expected to remember everything on demand.

[ThanksBAW]

Hope this helps!

[jshonw]

As mentioned already, the cap on the reset pin will interfere with debugWIRE.  But, it's good for the added protection against noise if you are not using it.  So, if I build it, I will just leave it out.

[McMonster]

And great work with expansion connectors, people can't really complain on the lack of modding possibilities. I just hope nobody will mention arduino shields now, there's simply no point

Sorry to dash your hopes, but I still think that to eliminate the chance to use a Ethernet interface is a big mistake. And the only most practical way of adding Ethernet is to use the $20 (delivered) shield.

I'm not against Ethernet shiled as a rule, it would be nice to have it, but I don't think it would be practical in this particular project. The design would have to be made around it, making space on the board for headers and arranging everything so that its' position makes sense with the case. By the way, wouldn't the new connectors be enough to make a simple adapter board for this shield?

[reagle]

Noticed the buttons did not seem to have any ESD protection around them- they just go directly to the IO expander if I remember correctly. I'd at least slap a small cap across each one and maybe a series element?

[EEVblog]

In what way the ethernet shield is ruled out?

Alexander.

Looking again, and you may be right. If the pin labelled SS is unused, then the Ethernet part of the Ethernet shield at least could be used. For some reason I thought that pin had been assigned to PS hardware, but I think I was wrong.

The SD memory enable pin (PD3) has been allocated for the rotary encoder. I guess if another spare pin can be found, you could edit the pin allocation in the library and regain access to the SD slot.

Is the SD memory enable pin actually required?
I was ensuring that all the SPI/SS pins were available to enable that Ethernet shield (or your own designed board) could be hacked in.

Does anyone know where the link to the new schematic PDF is? I tried the http://www.eevblog.com/uploads/ folder but only RevA is there right now.

It's there now.

Dave.

[EEVblog]

There's an AVR042 application note about hardware considerations for AVR chips

I previously pointed Dave to it, he didn't seem to be much interested. I was also surprised to hear his reasoning for changing the LED on the reset pin "after reading the datasheet". While this was in fact something pointed out be me in the forum https://www.eevblog.com/forum/index.php?topic=6163.msg80321#msg80321

Really that hard up for the recognition?  The statement certainly didn't imply that he found it after that, just that it was not something obvious and was "buried in the datasheet."

Several people in fact pointed out the RESET pin issue.
Sorry if you feel you didn't get recognition BAW.

Dave.

[firewalker]

I will be really easy to use any pin from the port expander under Arduino IDE.

Alexander.