[quote author="jason"]Yes, I have built and tested the filter. It worked as advertised, but not as drawn! Earlier when you mentioned I was missing feedback, I checked my notes and didn't know what you meant. However, I just looked at the schematic I drew in Altium Designer, and I see that when I drew the first 2nd order stage, I left out the connection between the inverting input and the output of the op amp! That mistake was duplicated on the remaining 7 stages in the schematic when I pasted it!
Anyway, thank you very much for catching that mistake and I'm sorry I didn't think to check the posted schematic rather than my notes earlier. Since this mistake is particularly grievous, I have edited my earlier posting to correct the error.[/quote] See if Ian can update the blog entry, too. Someone commented there before I had a chance. They caught the same error in the schematic.
[quote author="jason"][quote author="rsdio"]I must be misunderstanding your question about feedback. The filter itself requires no feedback beyond what is shown in the schematic to work. Each of the four 2nd-order sections produces a 12dB/octave lowpass response and the 3dB cutoff points are chosen in order to create the Bessel response. I used filter design software (Nuhertz) to verify the response of this filter
Are you perhaps referring to some other type of feedback in the circuit?[/quote][/quote] Have you actually built the circuit to verify that your schematic is correct?
Look at the Wikipedia entry for Sallen-Key. Figure 1 shows feedback from the output to the inverting (-) input of the op-amp. All op-amps require feedback for stability, unless you're designing a comparator, which the Sallen-Key is not.
Your schematic shows no link from output to input of any op-amp, thus you have no feedback. As mentioned in the blog comments, and op-amp will quickly slam up against either the positive or negative supply rails if you run without feedback, thus making your circuit fail with only DC output. Check.
[quote author="jason"]An 8th order lowpass Bessel filter meets the requirements I set out above, offering 40dB of attenuation at 20KHz, rapidly increasing above that frequency while maintaining a linear phase response within the passband.
Here is the schematic of the low pass filter. There are two identical filters in the design, one for the in-phase (real) channel, and the other for the quadrature (imaginary) channel. I have attached a PDF copy of the schematic for those who wish to examine it more closely.[/quote] It appears that you've implemented your 8th-order Bessel with 4 Sallen-Key 2nd-order filters in series. However, as noted in the blog, you have no feedback in the schematic, making it highly doubtful that the circuit shown in the schematic will work. Any chance you'll update the schematic or provide an explanation?
[quote author="ian"]To be honest, I'm not comfortable talking about these things to that large an audience yet. The readership here is a small fraction of the blog, so I can do things like this out of sight of most people :)[/quote] I think you've made the right choice so far.
Glad you had a chance to review this. Maybe I did not save the file after I set the 1 uH inductance.
What I would be most concerned about is modeling those ferrite beads correctly, because my intuition tells me they have non-zero inductance. Your results seemed to show the same current, and the rounding of the response did not really change the peak current, so perhaps the inductance does not necessarily need to be precise.
By the way, the 15 A or 16 A peak is probably what I got. When I said 50 A, I was going from memory, but as soon as I saw your graphs I remembered that it was only 16 A or less. I think maybe I tried another circuit variation that was up to 40 A, but that was probably a very old revision of the LTspice model.
[quote author="tayken"]In that tutorial, there is a IRFD110 (N-ch MOS-FET)driven directly by a uC pin, which is weird, usually you have to use some sort of driver IC or sth.[/quote] What do you mean? I drive the gate of FETs with a PIC I/O pin all the time. As long as the gate on threshold is positive, you don't need any interface, not even a current-limited resistor. The challenge is for an FET that requires a negative gate voltage.
EDIT: I just grabbed the IRFD110 data sheet, and VGS(th) is minimum 2 V, maximum 4 V, so as long as your uC is running 5 V I/O outputs then you should have no problem driving the Power MOSFET directly.
[quote author="tayken"]Thanks Ian, I guess we can put a series resistor on ADC pin to get some sort of feedback and run a closed loop for adjusting the duty cycle accordingly.[/quote] Do you plan on voltages over 5 V ? If so, then you'll need a voltage divider instead of just a series resistor. Also, if VUSB drops below 5 V then the ADC input won't accept 5 V, so you might still need the divider.
The AP2141 is an interesting part, so thanks for that, but it represents an addition to my existing circuit. Since I can solve the problem simply by replacing the 100 µF and 47 µF caps with 1 µF and 0.68 µF caps, I think I'll go with the cheaper solution with fewer parts to break down.
[quote author="ian"]@rsdio - Yes, that is another example I repeat frequently :) It's not just garage assembly lines and internet startups, established businesses find it overwhelming too.[/quote] Meanwhile, my new startup is launching its first hardware product (after a software product that helped raise funding for the hardware R&D). All work that can be outsourced is actually being kept within the local area - within the same county. The only outside sourcing is the parts, which are obviously manufactured in Thailand, Philippines, Japan, and all over SE Asia / Oceania. We have a small factory for final assembly, but at my urging we are hiring professionals to do every task that can be done as a "standard" service. That way we can spend all of our time on the unique parts of our product design instead of losing all our time doing the same things that other companies are doing. We've already worked with two local PCB fab houses, and there are more hungry local ones to choose from. We've also worked with two local assembly shops, one with better pricing on small runs of 1 to 4 boards and another with better pricing on runs of 100 boards. Again, there are more hungry local shops to choose from if we take the time to request more quotes. These places are all close enough that I can drive to pick up the orders when they're complete (in my electric car!) without waiting for UPS or FedEx to deliver, and one is even close enough that I can walk!
I realize that not every hardware designer lives in an area where there is so much competition for this kind of work, but sometimes you'd never know what sorts of businesses are hiding down some industrial road that you never see. Many of these places are not even on the web or listed in the phone book. One of them even acted very surprised when I called them on the phone, because I think they assumed I was selling something instead of actually wanting to hire their services. But, once I started working with one or two of them, I started learning about the rest. The PCB fab houses will tell me which assembly shops are hungry for business, and which ones have a reputation for good work. Meanwhile, the assembly shops can tell me who their customers use for PCB fab, and who has the best pricing. The assembly shops are also answering many of my questions about design for manufacturing, giving advice on resizing those default Eagle package pads to work better with SMD ovens and their particular P&P.
Bottom line: Look around your city and see what you can learn!
[quote author="ian"]it is only intended to be a programming voltage (pic uses 1uA or something small of 9-13v), bias voltage, or reference, not a serious current source.[/quote] I assume you mean that the programmed part is not a serious current load, or current sink, right?
I don't think that a 9 V to 13 V "reference" would be appropriate (overkill?), but a regulator or any other voltage should be adequate.
[quote author="bearmos"]I'll be curious to see the simulation if you get it uploaded (I also use LTSPICE).[/quote] Sure thing! [attachment=0] When you run the Simulation in LTspice, use the Transient command with Stop Time: 100us, Time to Start Saving Data: 0, Start external DC supply voltages at 0V: "checked" The actual command should read ".tran 0 100us 0 startup"
You should see the current rise to 450 mA in a few microseconds, then hold at 450 mA until 20 µs, after which it drops to 0. In order to see this, you'll have to click on D1, L1 or L2 so that the current is graphed.
If you change the 1 µF cap to 100 µF and the .68 µF cap to 47 µF, then you'll see the current spike to almost 50 Amps!
Notes on the circuit.
D1 allows me to mix power from a 5 V wall wart via another diode that is not shown (and also no longer used or populated).
C1 and C2 are placed near the 3.3 V and 1.2 V dual-voltage regulator input pin. C3 is placed near the 5 V boost regulator input pin. R1 simulates the load (I actually measured 1.04 MΩ in my circuit, but I don't know how realistic that is).
L1, L2, D2 and C4 all come from a Spectrum Digital (Texas Instruments) schematic. The Zener protects the circuit from a faulty USB power source that exceeds 6.2 V for any reason. C4 supposedly removes USB Host power supply noise (some of it, at least). L1 and L2 are the correct part on my boards, but I don't know what the correct simulation should be, since the part's data sheet gives no value for inductance but the LTspice model demands that I type in something (I used 1 µH and have not actually had time to test whether other values change the results significantly - that's an important step that I should take!).
C5 came from some other circuit that I saw, and I'm not really sure whether it's redundant. I suppose a little extra supply filtering should not hurt, at least not at .1 µF
Any comments at all are welcome ... or questions if you can't get the Simulation to run or if the file is somehow damaged. I use OSX for nearly everything, except I use XP for MPLAB, CCS, and SPICE, so I had to ZIP the LTspice file and ftp it to my main machine for uploading here. I hope that worked!
[quote author="tayken"]Can't we do with 5V USB input? Also what do you think about current requirements?[/quote] USB delivers at least 4.01 V, guaranteed, and that voltage should be enough to generate any voltage using an inductor and switching boost regulator. Problem is that the power would be limited to well under 2.5 W, but I guess that's understood.
I wonder whether a DAC output could be mixed in the feedback loop of a boost regulator to allow PIC control of boost output voltage. There should be an example circuit somewhere...
[quote author="ian"]We could go P&P, but that is a whole other can of worms. In my open hardware manufacturing presentation I always take a second to admonish people not to buy pick and place machine and start a garage assembly line. It's a cool machine, but it eats up 6months to a year of core profit activities to get it setup, and you only get an extra few cents per board for that investment. I know now of dozens of people world wide who went this route and became a slave to the P&P instead of building their business, and lost a ton of momentum. Every time I give this presentation someone tells me afterwards that they fell into this trap too. Adafruit likely had millions in revenue before buying a P&P, and I don't know how much they actually use it. Seeed bought a P&P (multimillions in revenue a year), and Eric says they don't really use it because it's faster, cheaper, and cleaner to send it to a pro assembly house. SparkFun has 2 P&Ps @ 32million revenue, Bert and Ernie, and their buyer (Pete, I think) also told me they are sending things out whenever possible instead of using their P&Ps. We're riding at just about (maybe we break it?) $1million in revenue this year, until we get to $10 I wouldn't seriously consider a P&P, even then...[/quote] I've mentioned this elsewhere in the forum, but it seems pertinent to tell the story again here. One of my commercial clients in the USA got tired of communications problems with China and decided to buy their own P&P. They never had an "Eric" like Ian has at Seeed, so every new project resulted in 100 non-working boards and lots of money lost on parts and shipping. Sometimes TSA or Homeland Security kept PCBs at the border for months. They got a 6-figure bank loan, bought a P&P with enough feeders to make their complex PIC-based USB board, and then slowly learned that they did not have the skills to operate a robot placing 64-pin TQFP packages. After all that expense, they now outsource board stuffing to a local business in the same city (that they could have easily used without a huge bank loan) and use their over-qualified P&P to place buttons on simple front panel boards.
In other words, I totally agree with your presentation.
[quote author="BrentBXR"]I dont understand why you dont find a differnt factory, closer to you. and purchase them yourself and sell through your site? Why do you middle man with SEEED? Or you should atleast have like half the order shipped to you; then sell that half to USA customers and asia customers can buy from SEEED or something. I would like that![/quote] These are all open designs. If you think it's a good idea to manufacture them in the USA for direct sales to customers, then why don't you try it? You'll quickly understand how difficult it is to avoid losing money all the time.
However, you should never underestimate the costs (time & money) of running your own business, putting things in boxes, dealing with shipping concerns and internet orders every single day, not to mention the serious government regulations and taxes. Frankly, if Seeed is able to handle all of the boring stuff, then I totally understand why Ian is not doing any of this himself.
[quote author="ian"]The bottom line, I think an eventual update of the Bus Pirate will need a fet and coil for a SMPS, and a way to switch it :)[/quote] I just finished successfully designing and building a USB board with the MAX1797 configured for 5 V output. It has an integrated FET, so there's no need to add an external one. It has a shutdown input that can be driven directly by a 5 V I/O pin, or if you only have 3.3 V I/O then this chip will still work. It has a Low Battery Indicator input/output that can be looped back into the Shutdown input, so you can feed 3.3 V logic to the LBI, attach a pull-up to the LBO-to-SHDN node, and you're good to go. I'm assuming that you only need 5 V from the SMPS.
There are two variations of the MAX1797 (MAX1796, MAX1795) that carry less current and also cost less.
I used the shielded SMD Abracon ASPI-0428S-220M-T for the power inductor; an alternate is the Sumida CDRH4D28NP-220NC.
