I had hoped to to have completed the basic design by now but I've spent a lot of today thinking about
power supplies, but not getting quite far enough to commit to paper (or CAD).
The idea was always to have a a separate power conversion board, so there is no reason that the rest
of the design can't be offered up for discussion without the psus.
All the schematics will be posted as .pdf s but it's probably easier to have the blurry .jpgs in this blog for
instant reference.
I've added a second sigma delta converter to the design so it has the ability to accurately measure the
output current. At the expense of yet another relay it could also measure output voltage. The reasons for
doing this is so that those who want can use voltage control of the output, and those who want to measure
complex load impedances are able to do so.
A small digression is necessary here to explain my thinking, because it affects other decisions as well.
My intention with this design is not to end up with a commercial product but something that can be made
and be useful but is also capable of being re-purposed and tinkered with. With that in mind it already
has optional things that you can leave out and far more support for debugging and expansion than
most commercial devices. It must also be admitted that it is not designed to a commercial standard,
and will probably require tinkering ! Although I have spent quite a lot of time on the design it is only
a small fraction of the time and testing that I would expect to devote to a paid for design for production.
The ADC and DAC section has grown an extra ADC, as discussed above, but is otherwise very little different.
The second ADC is driven by an amplifier running from dual supplies so it could drive the ADC input negative
which might do it harm - so it has to be protected by D21 and D22.
For a more complete description check out the earlier blog about the ADC, DAC and FPGA sheet.
This is one of those dull schematics that could almost be a spreadsheet, but there is a little bit of stuff
worth pointing out.
The processor has two external flash memories, U17 which is 4kbits and for calibration data and U15 which will be
64 or 128 Mbits and is for data. I've found, through bitter experience, that it pays to keep them separate.
U16 is the FTDI USB interface chip which looks like a COM port to a PC that connects to it.
U13 is a precision voltage reference for the micros ADC and DAC. U14 is a a reset chip which will reset the
processor if the battery voltage is too low.
J6 is the connector to the Riverdi display - its connections and some extras are duplicated on PM1 which is a
PMOD compatible connector which can be used for a different display if required.
There are three more PMOD connectors, two have access to processor serial ports and are intended for Bluetooth,
expansion or whatever. PM4 is connected to otherwise unused pins on the FPGA (actually one of the pins is shared
with the user control connector but I shall try to avoid connecting anything to it.)
Finally we have the output section.
This has taken more work and thought than I expected when this project started.
Dave Martin has been urging from the start that we should provide more drive voltage than was my original plan
and I was already thinking along those lines when he posted his useful measured data the other day. This data
makes it clear that more volts would be a good idea.
This new design goes as far as we can without excessive cost or risk of electric shock.
I'm using OPA551 amplifiers which can work with 60V maximum Vspan and deliver a maximum of 200mA (more like 100mA in real life.)
The design of the current source has changed from my earlier efforts and is significantly improved. In the original design both amplifiers
were connected as current sources operating in opposite directions. This required 4 precision resistors per amplifier and still had
problems with balance and offsets.
In the new design U19 is configured as a current source but U20 is configured as a unity gain inverter so that its ouput voltage is
equal to but opposite in polarity to U19's. U19 controls the current but U20 doubles the voltage compliance - but U20 doesn't need
any precision resistors.
There's another twist which gets rid of offsets in the current which can be caused by errors in the resistors around U19 - C69 AC
couples the feedback which makes U19 a current source - which reduces the DC offsets to minimal amounts.
If you really want to do DC coupled work, then link out C69, or better still, remove C69 and reduce the value of R75. (In current
source mode R75 = R73, so if you open the R76/C69 arm of the bridge U19 works as a unity gain inverter. If you then reduce R75 to
R73/20 U19 will have a gain of 20 and the input voltage that would have given you 10mA will now give you 20V, but U20 will drive the
other output to -20V , giving 40V across the terminals.)
The maximum output the OPA551s will manage is about +/23V, so we can get about 44V pk (and 88V pk-pk) into the terminals
once the drops across the resistors R74, R79 and R82 are taken into account.
PTC1 and PTC 2 are protection devices which will limit the current flowing into the protection diodes on the amplifier outputs if
a large external voltage is applied. The PTCs work by getting hot, which takes time, so you still need resistors to limit the maximum
current. ZD1 and ZD2 are there to stop the external fault current from pulling the power supply voltage too high.
R82 is a current sense resistor and the voltage across it is measured by U21. The INA149 amplifier shown is very expensive (about £7)
but it does have a good enough common mode rejection to measure the current accurately. There are cheaper high input voltage
difference amplifiers but they won't be very accurate. RL4 allows you, at the expense of losing half the voltage on the output, to get rid
of the common mode voltage on the current sense resistor - and then you can use a cheap part for U21.
U22 provides a half supply reference voltage for U21 and U223 provides the necessary complementary signal to drive the ADC.
It's a lot of bits and a fair amount of expense to measure the output current properly - you could decide not to bother and for
some applications this will probably be fine - but we won't really know until we try it, (For impedance measurement accurate source
measuring is essential.)
This is the pcb from the component side, you can just make out the display in yellow on the other side. I won't start the
layout until there's been some discussion (and I add in the bits I've forgotten.) There is a prize of infinite Kudos (but nothing
else) to whoever spots the missing things !!
There's quite a bit to digest there I'm happy t to answer questions and receive suggestions.
MK
Top Comments