Community Portal

From Mozda Converter

Current revision as of 09:30, 22 October 2007

Lets Discuss on this page the important questions, jot down our thoughts, divide the work & post Minutes of meetings.

Status & MOMs

Meeting on April 8th Sunday

We discussed pros and cons of Tim Nolan's and Tunecharger circuit

====Tim Nolan's ckt==== is a buck converter, has too many sensors (V,I at input output) and uses a bigger microcontroller, making it more expensive. ~100$ for bare bones prototype. It implements end of charge detection by hardcoded o/p voltage threshold. Tested at 100W,20V,5A input. Inductor rated at 11A, Mosfet at 50A. So it can handle 200W. Thus for a pedal setup, it is limited by voltage. For higher voltage following parts will need to be changed. Input cap rating(25V), Mosfet G-S limit(20V this circuit uses input to create this difference, which can exceed rating when input is higher. Tim's suggestion: modify bootstrap circuit to use voltage regulater like 78M15 before diode D4 to limit Vgs). Out of the sensors, only output current sensor is needed for MPPT, and o/p voltage sensor is needed for end of charge detection.

• Pros: only needs to be modified for higher voltage, not wattage.has easy but hardcoded end-of-charge detection.
• Cons: Since both V & I sensors needed, cost around 70-100$.

====Tunecharger ckt==== Is based on boost converter design with a discharge mosfet in load circuit, thus charging the load in a pulsed manner. Circuit is simple, can support both stepup & down, has in built MPPT with only voltage measurement. But needs extra temperature sensors for end of charge detection. (based on mosfet/battery overheating). It has been tested for wind/solar/battery sources for upto 20-30W. HIgher wattage & highervoltage can cause ratings to exceed. Inventor is not ready to disclose 50W ckt in developement.

• Pros: can charge from wind/solar/pedal. Only V sensor needed for MPPT, smaller microcontroller. So less cost ~30-50$.
• Cons: not tested for high V & wattage yet. There might be deterioration of battery performance due to pulse charging. Need temp sensors for end of charge detection.

Questions

A lot of questions came up in today's meeting, which need further study of circuit for resolution. These are as follows

• Tunecharger converter being of boost type inherently, it can only do limited stepdown (2.5 max). Hence we either have to append a buck converter at input or increase the load from 12V to around 24-48 volts for charging it with pedal generator operating at 48-70V. Michael is already using 48V battery bank in Mozda grid. But for more general implementation ability to have 12V load might be more useful.ToDo: Ask Michael, if 40-70V to 24/48V battery converter is useful for him.
• Overpowering Tunecharger: Need better grasp of L & C sizing for tunecharger operation, and various bottlenecks in current design which might be reached for high power/voltage operation. Since some people on newsgroup reported mosfet heating and loss in efficiency.
• Efficiency analysis: there is some ambiguity about theoretical analysis of average power in tunecharger being optimal or otherwise. To Do: study tunecharger operation & conduct thought experiments to resolve this quickly.
• Can we modify the tunecharger design to make it a buck converter while maintaining simplicity of V sensing for MPPT?

But we have decided to go ahead with basic tunecharger implementation to test efficiency/ proof of concept. It can further be modified to suit our needs. Achintya will order the parts. Ameet will buy a big enough breadboard. Ameet will also find availability of suitable labs in MAE & ECE department, which have oscilloscopes, variable voltage sources & other prototyping resources. (pcb fabrication !)

Meeting on 22 March Thursday

• We decided to start building the prototype.
• But there are two competing designs to look at. Tim Nolan's & the one at tunecharger website.
• Both the designs are not currently designed for our power & voltage requirements. Hence first task is rudimentary redesign then prototype building, troubleshooting & final design.
• Benefits of Tim Nolan's ckt is that it is a buck converter, with similar step down ratios like ours. The initial parts cost seems to be within 100$. Tune charger ckt is a flyback converter, enabling voltage stepup as well as stepdown. It implements pulse charging. Initial parts cost for low voltage is around 50$. If successfull, it can work for wind, pedal, solar etc. One problem of tunecharger design is the high voltage on the capacitor. But clearly it will take some time to see through obvious bottlenecks & design something which will work for our design. (efficiency+MPPT+input side variability+various output voltages)

To Do:

• Study both designs.
• Design modification for pdeal power
• Design for efficiency, cost, MPPT, input/output variability in that order.
• Brainstorm what can go wrong, to foresee the pitfalls.
• Build prototype.
• Next meeting: sometime next week.

Meeting on 9 Dec Saturday.

Ameet, Rathna, Abhijeet & Achintya met yesterday.

ToDo:

• Rathna & Abhijeet to work on finding cause of inefficiency in UIUC design. This will try to do pspice simulation of UIUC circuit, to check if the simulated efficiency was itself low. They will also study the datasheet of the MOSFET used, and if its switching losses can explain the inefficiency. Pspice student version is free, but has freq limitations, and limited part library, which can halt the progress.
• Both Rathna & Abhijeet has friends in Power Electronics industry. They will inquire the feasibility of such our design, and how to go about building it.
• Achintya is studying converters and various types. He is also studying Hugh Piggot's plans. Achintya, could you work on determining if the conduction losses before the converter stage will increase/decrease by stepping up the voltage?
• Abhijeet will try to get hold of pspice/similar license once Himanshu is back.
• Ameet will try to understand the existing circuit, and how can it be modified for higher voltage application, by consulting with Rahul Kodkani. Also to mail all relevent links, to the team. Construct webpage on Udai for this effort. (creating this wiki instead)

Discussion

• Questions we faced are listed on Q&A page
• We figured from very rough calculations that around 70% efficient converter is needed to match efficiencies of current system operating at 48V without the converter, which has 20W input side conduction losses.

Saurabh: I'm not sure I understand this yet. Are you trying to say that if we are to waste at most 20W of power when generating at 150V, then we should be using a converter with >70% efficiency? And is this based completely on Hugh's "claim" that resistance increases as square of voltage in any generator? Or is there something else going on?

Ameet: No.. no. It is not based on hugh's claim about resistace. It just says that, in current system we are wasting 20W out of 75W for conduction losses. In the new system, since we are increasing V 3times, we will cut back current to 1/3rd and losses will be 1/9th. (not assuming resistance increase, which really will)So we have 3W of losses. To get more than 75-20=55W out of 75-3=72W, we need to have converter efficiency greater than 55/72=76%

• MOSFET's switching losses depend on switching freq, which is PWM frequency. Its conduction losses are negligible.
• Too many Diodes in the circuit. What is the need? Diode reverse conduction losses negligible in schottky diode. What explains the converter losses then?

Achintya: The number of diodes is necessary since the AC current comes in 3-phases, so each of these phases needs to be independently rectified.

Ameet: yes. but extra diodes I was talking about are after rectification stage, in the buck converter

Next Meeting Meeting time changed. I think it is on tuesday 5pm. Rathna shall send mail about the same.

Questions for Michael

Invited Michael to join our Wiki, and asked a quite a few questions. Here is a transcript.

Right now we are involved in evaluating if there is any benefit expected from steeping up the voltage to 150V. This involves whole system understanding. So have a LOT of questions :-) It would be great if you can answer them. Some things might be very basic, but we dont want to go ahead with wrong assumptions. Here they go..

• What exactly does the current system with 20W losses look like? Is it generator directly connected to 4, 12V battery bank? or is it connected through charge controller? (which one? so that we can search for datasheet). I think that all charge controllers have Dc-Dc converters. So why design new one? (I am guessing lesser cost) Also what is load resistance/ battery bank resistance?

Michael: There is no current PPG system. We have a 48 volt windgenerator, and we have some data for it too. But we do not have internal loss data. However, we can make reasonable estimates based on various measured/observed figures. The 20 W loss is an estimate by Hugh Piggott, for a coil design I am testing now, based on his suggestion. The windgenerator does charge a 12V X 4 batteries battery bank. However, this is not going to be the ppg system. It does not charge the battery bank through a charge controller. Charging is done directly from the windgen using bridge rectifiers, there is a shunt regulator on the batteries, which diverts the windgen power to a dump load when the batteries are fully charged.

We want a newly designed charge controller because there are no cheap MPPT controllers on the market, cheapest is 150$ for 12V, 250 watts, and no one will vouch for it. We want to minimise losses in the stator, that is why we want to go through this exercise.

All charge controllers do not have a DC-DC controller built in. Only MPPT charge controllers do. The rest are either series controller (which connect or disconnect the power source depending on state of charge of battery) or shunt regulators which divert the power from source to a dump load or short the power source (as the Steca controllers do, you can see that at www.steca.de I think).

• What is winding resistance of generator? How did you figure out 20W losses? (I am guessing, V*I. What are the steady state current & voltage of generator & pedaling rpm at that speed)

Michael: Winding resistance for one phase of our windgenerator is about 1.4 to 1.6 ohms (which is 18 SWG wire, 378 turns per phase). The resistance per phase of my test coils for ppg is 0.4 ohms. I tested at 275 rpm, 58.5 watt output per phase, this gives about 14 watts of losses inside the stator. I still need to test this system at 225 rpm.

• Right now, generator is labeled as 48V machine. Is this because it always produces 48V, or it is a nominal voltage. I am guessing that as pedaling speed changes, Voltage will change too. And torque changes as per current. So in that case, would 48V be full load voltage/ nominal voltage?

Michael: 48V is the nominal voltage. You are right about voltage variability and current being proportional to torque.

--Ameet 23:49, 11 December 2006 (EST)

Questions & Answers

Right now key questions to answer are (lets append the answers alongwith as we find them)

• what is cause of low efficiencies in UIUC design. Is that MOSFET? poor design? could they not catch it in pspice simulation they carried out before the design?
• Are there theoretical benefits of higher voltage? There are conflicting opinions on this, as Hugh Piggot thinks R will increase as square of voltage, nullifying the benefits, while some others think it is good idea. Can we analyze this more carefully to get a final answer?

Some not so key questions are

• saurabh: why not do ac-ac stepdown by transformer? more losses?

Achintya: I mentioned something on this in the blackbord section, but reading through the DC-DC converter manuals, the impression that I got was that AC-AC converters are actually more efficient ~90%, and definately easier to design. The cost would be the thing to check

• ameet: what is Michael using currently to down convert 48V to 12V? I did not see any converter in the schematic. I just saw charge controller for one of the batteries. Rest of 4 batteries were connected in series tto make 48V.
• ameet:How does generator voltage stay constant at 48V? does it not fluctuate with wind speed? Answer should lie in Hugh's Wind Turbine handbook.
• rathna: what efficiencies will make the converter more attractive than current setup?e.g. If we are saving 15W in input side losses, and lose those in converter losses, we better not use it :-)

Blackboard

Use this to jot down your thoughts/comments/suggestions along with your Name.

• Saurabh mentioned something in his e-mail that I was thinking about too. Why not use the AC current at a high voltage directly from the Axial-Flux generator as the main transmission signal and then use a simple step down transformer and rectifier close to the battery? The cost for the MPPTs scales with voltage and considering that the power being transmitted, ~70 W, is not that high, it does not make sense for Michael to buy the outback MPPT when there are much cheaper ones that deal with lower voltages. The ones that are able to work with volatges of range ~12-24 VDC are perfect for the application. The only snag in this suggestion is that the current directly from the axial flux generator comes in various phases depending on the ratio of the number of coils to the number of permenant magnets. My guess is that the output is a 3-phase AC current and we would need 3 pairs of wires as opposed to 1 pair to carry this current from the bike to the battery pack. I dont know if there are any inductive losses assosiated with having 3 out of phase currents travelling in a bunched wire pack, but this is definately something to ask Michael about. Also, I didn't check on the cost of trasformers from say 160 VAC - 15 VAC, and since we'd need three of them (for each phase), I dont know if this is less effective cost wise. This is something to ask Michael about. - Achintya (12/11/06 - 4:30pm PST)

I think, given high voltage this might make sense. can you follow up more on it, and present at next meeting? Lets discuss it within ourselves before pitching to Michael. Most important thing is efficiency! I used to think, transformers cant have more than 70-80% efficiency due to losses. Another thing is. We dont need MPPT. They are designed for photovoltaics to track maximal power tracking point. We can do without that intelligent feature. All we need is a user settable Dc-Dc converter. Also number of phases in generator is 10 :-) as there are 10 coils. They might be connected to get 3 phase, but I am not sure. I think Hugh does discuss this issue somewhere in his manual.--Ameet 19:20, 11 December 2006 (EST)

"Hello, I am Jean-Michel Cour, the inventor of the TuneCharger. I think that most issues discussed here are now solved the our 4th generation of battery chargers. The charger logic will soon be marketed at low price as a preprogrammed microcontroller that allows customizing to your power needs. The new logic has buck and boost capability, and end-of-charge monitoring as well. My email is jmcour[at]tunecharger.com. Hoping to read from your community soon ! "