AC regulation and active rectification circuit

[ACharnley]

Have some questions? What current on the 5 volt output?
How much current is available from the generator at (full speed) and (at low speed)?
Is this a cell phone charger?

Using Mr SPEC's idea:
Using a voltage doubler to get 2x higher voltage.
Using a 4 to 140 volt PWM. at 400mA
Will need a change so the PWM will output less current at low input voltage. (easy)
View attachment 96266

The dynamo is 6W/6V unsaturated, so it should give just over 1A. This depends on speed and load, though other people's tests show 12-15W at higher speeds being plausible.

So for the circuitry, at least 1A, ideally with scope for a bit more. 400mA wouldn't be enough to charge a smartphone. I looked for a PWM that would cover 100v a few weeks ago but they all began with around a 7.5v input voltage (and offered low current).

Also at these voltages Schottky diodes would break down, so back to a 0.7v drop?

 

[spec]

From what I now read, synchronous rectification is another term for swich mode voltage regulation, which is already sorted (not part of this circuit). This circuit must a) step the AC down to AC below 20V and b) rectify it to DC as efficiently as possible.

Hi Andrew,

As has been stated, 'switch mode' and 'synchronous rectification' are two independent and separate techniques.

(1) Switch Mode

If you have a supply voltage, say from a rectifier and reservoir capacitor, of say 100V and you want to generate a 5V stabilised supply, you have to regulate. Note that the voltage from the rectifier will not only vary with changes in the alternator speed, but also with temperature. On top of that there will be a ripple voltage saw-tooth which will be inversely proportional to the size of reservoir capacitor and alternator speed, and directly proportional to current drawn from the reservoir capacitor. In short, the voltage on the reservoir capacitor will be all over the place. To get a stabilised 5V supply you have three options:

(1.1) Series linear regulator

Here you have a control element in series with the target output voltage. Its job is to waste the voltage difference between the reservoir capacitor and output voltage. it does this dynamically so that responds to any changes, including ripple voltage, on the res cap.

This is by far the best and simplest way of doing the job, with one reservation- it is dreadfully inefficient and simply wastes the power necessary to maintain a constant output voltage.

Take the example: 100V in 5V out. Say that the 5V supply is providing the full USB current of 1A. That means that the series regulator must waste Vin-Vou *I= 95 Watts and there is the problem. (of course in the case of the cycle alternator this would not happen but don't worry about that for the explanation)

(1.2) Shunt linear regulator

The shunt regulator does the same job but it would dissipate 95W all the time, 1A load or 0A load.

(1.3) Switch mode regulator

The switch mode regulator overcomes this massive power loss and but operates in a much more complex way. You have 100V and you want to generate 5V at from 0A to 1A say.

The Switch Mode Power Supply (SMPS) first charges up an inductor.

VL * T = IL. The energy in an inductor = (IL ↑ 2)L

When you put a voltage across an Inductor the current through the inductor will increase literary like a ramp.

At a time defined by design the, inductor is then connected to a capacitor across the 5V supply and some of the energy stored in the inductor is transferred to that capacitor. As a consequence the capacitor voltage rises. When it reaches 5V, the control circuit switches the inductor back to the res cap. This switching takes place typically from 1K Hz to 4MHz, depending on many design parameters

The net result is that the 5V supply is stabilised and unlike the linear regulators no power is wasted.

Of course, in the real word some power will be lost but not much. SMPS are typically 80 to 95% efficient. Assuming 90% efficiency, this means that only 5*1*0.1 =500mW would be dissipated, a bit different from 95W of the two linear regulators!

Note that I have greatly exaggerated and simplified to illustrate the principle.

(2) Synchronous Rectification

Before the widespread use of SMPs, synchronous rectification (detection) was mostly used for high-speed radio frequency detectors, simply because it can be made faster than a simple diode detector (much simplified).

Before describing the operation of synchronous detectors in SMPs, it is best to recall what is going on with a conventional mains transformer,diode, capacitor (TDC) rectifier.

(2.1) Transformer, Diode, Capacitor Rectifier

The first point to note is that it is amazing that the transformer capacitor diode rectifier works at all, let alone reliably. The only reason it is used so widely is because it is cheap small and easy to design. It only works so well because the manufactures have done a fantastic job of making diodes that will stand the abuse they are exposed to every cycle of the mains supply.

In a conventional rectifier circuit the diode only conducts for a few milliseconds at the peak of the AC sine wave from the transformer secondary. This means it has to pass a massive current for a short time. Consequently the instantaneous dissipation is very high, not to mention the other stresses the diode suffers.

(2.2) Rectifying Element

Taking a humble 1A 1N4001 rectifier diode, the data sheet shows that it can pass 1A average current, but its peak repetitive current is 30A. At 1A the diode drops 1V1 and at 30A, 2V. 30A would not be a very high peak current in a rectifier circuit but, all the same, the 1N4001 would have an instantaneous dissipation of 30W. Not only does this mess up the stability of the DC rectified voltage and stress the diode, but it is also a total loss and even though it only lasts for a few milliseconds, the average energy loss is significant. This is one of the reasons why the voltage of a TDC rectifier sags under load.

A very basic model of a normal silicon rec diode is a 600mv battery (band gap) in series with a small resistor. Shockey diodes improves on this with a bandgap of 400mV.

A fully turned on MOFET is different. Between its drain and source it is a resistance, there is no band gap. A high-power MOSFET would typically have a drain source resistance (Rds) of 15m Ohms at 25 deg C. That would rise to around 30m Ohms at 100 deg C, a typical junction temp for a rectifier. Taking the 1N4001 example with an instantaneous peak current of 30A, that would be a voltage drop of 30m Ohms * 30A= 900mV, quite a bit lower than the 2V of the 1N4001.

You could not simply replace a diode with a MOSFET. Instead you must turn it on just before the peak of the Tx secondary sine wave, and off just after the peak. In fact it is synchronised with the mains frequency, Hence the term synchronous rectification'. '

The other advantage of synchronous rectification is speed. The turn on and turn off can be made faster than a diode so switching losses are reduced, but that is a whole story in itself!

 

[ronsimpson]

I made some changes.
Added two 68V Zeners on the input. This will limit the voltage to 136V approx. Maybe the next lower voltage part should be used.
D2,3 are Schottky to have lower loss. (150V diode)
Increased the value of L1 and C1.
There are two smaller IC if the current is low. Or R2=100k or 50k to reduce the current.

This circuit has some funny things. The PWM can output 0.4A. The input side is 0.2A at 11V, 0.02A at 110V.
So at high speed it pulls very little current but a low speed it pulls hard. Because it is designed to draw constant power.
That might be a problem.
For more current, make two PWM and run them in parallel. (0.8A)
There is a way to make it not work until the input voltage reaches some voltage. Set by two resistors. Now it will try to work at 4Vdc. Which is about 3Vac from the generator.

 

[ronsimpson]

The dynamo is 6W/6V unsaturated, so it should give just over 1A.

In the voltage doubler mode it will deliver 12V at 0.5A-(diode drop). Then a PWM could get you to 5V at 1.1A. approx.
If you used three of the LTC7138 then that's 1.2A. You could set the first PWM to start at 4Vdc, the second to start at 4.5V and the last on to start at 5V. If you want to not pull hard at low speed.

 

[spec]

Looking good Ron,

I think a better characterisation of the cycle alternator is required. I would like to see a graph of output volts versus expected RPM range when it is feeding the rectifiyer cct with say a 5K6 resistor across it. I would be amazed if it reached anything like 200V

Any idea how many poles the alternator would have: 2? My old bike, so called dynamo, had 2.

 

[ronsimpson]

STOP. Much better idea coming soon. Need to go make $$$s.
Need to know what is the slowest RPM of the generator.
How many poles?

 

[ACharnley]

Some cracking responses here!

The dynamo is 18 poles. Things have moved on in the cycle world!

There is a video of a Shimano hub on youtube being driven by a drill, and it makes just over 100v, however this is with no real load on it.

Unsure of the Hz, but it's real low, 4Hz or something at minimal speed. I'm not expecting anything to work at this speed. Probably 20Hz will be the minimal before turn on.

I'm still seeing loses at the 6V range through the use of the diodes, which is what I'm ideally trying to avoid (at the expense of high speed efficiency). Perhaps the mosfets could be used for regulation before going into a switch mode with a higher voltage rating, rather than relying on the darlington's to dump it as heat, but I can't find one that'll do it and work at 6V.

 

[spec]

Hi Allen,

Can you clarify what you mean by seeing losses caused by the diodes at 6V. Can you post a schematic showing exactly what you are doing to establish that. Not wise just yet to assume you wont get an output at low RPM.

 

[alec_t]

If you want the FETs to operate below 6V or so you'll need to be picky and find some which turn on fully with a really low Vgs.

 

[spec]

Nothing Ron can't handle

Using a bipolar tran would be one way

 

[spec]

Some cracking responses here!

The dynamo is 18 poles. Things have moved on in the cycle world!

There is a video of a Shimano hub on youtube being driven by a drill, and it makes just over 100v, however this is with no real load on it.

Unsure of the Hz, but it's real low, 4Hz or something at minimal speed. I'm not expecting anything to work at this speed. Probably 20Hz will be the minimal before turn on.

I'm still seeing loses at the 6V range through the use of the diodes, which is what I'm ideally trying to avoid (at the expense of high speed efficiency). Perhaps the mosfets could be used for regulation before going into a switch mode with a higher voltage rating, rather than relying on the darlington's to dump it as heat, but I can't find one that'll do it and work at 6V.

Not much; I had many bikes with N pole dyno hubs. I was talking about the 'dynamo' type that rubs on the tyre- you can still get them.

Yeah, permaneant magnet alternators will generate an infinate voltage off load, in theory that is. If the energy in the alternator coil has no where to go that is what will happen.

Any chance of doing the test that I mentioned? I feel sure that we our chasing our tails with this 100V thing.

Don't concern yourself with diode drops- they are not an issue. Who cares about the odd volt here or there. The real issue in the alternator energy output.

 

[ronsimpson]

I need to tell a story and see if you all agree.
Years ago I designed, something like this, for a large wind generator.
The current was (mega) and the spec said the voltage was 3,000 volts. Why 3kV?
In the case of a hurricane, AND the power grid went down so there was no place to dump the current, and the AND the 'lock the blades' function also failed, the generator would run-a-way and make 3kv.
So I made electronics that worked at high current under a short. AND In a over voltage condition it dumped heat into the steel tower. This kept the voltage to less than 1kV. So I did not design for 3kv because I made a load that kept the voltage down. My design had less parts, more efficient, low cost, less failure. The design was rejected because it did not meet specifications. The design worked well under a run-a-way condition.

SO: I think the alternator will not go to 100V with a 1 watt load. Probably not 50 volts. I can add a 2 watt resistor and small transistor etc that will not function below 50V but shifts in at 50V. We need to learn more about what 1 watt of load does at full speed!

The dynamo is 18 poles

That means the frequency is 9 or 18 times the RPMs. (?)
Is the dynamo connected to the hub so one turn of the tire=one turn of the alternator?

 

[ronsimpson]

Unsure of the Hz, but it's real low, 4Hz or something at minimal speed. I'm not expecting anything to work at this speed. Probably 20Hz will be the minimal before turn on.

The size of the capacitor in the voltage doubler is very dependent on the frequency. 20hz is much lower than I thought.

 

[spec]

Like the spec bit- same here many times. Suits wont listen!

I have been saying all along that I didnt think that the alternator would go to 100V on load. Anyway the answer to all this is to fit a very large capacitor- like a LiIon battery.

 

[ronsimpson]

I think D5,D6 = 12V 1 watt Zeners and will limit the voltage to 24 volts. I can not prove it with out a generator but I think this allows us to use low voltage parts. I think anytime there is even 1 watt of load the Zeners will not pull current.
Here is a 6 to 30V PWM that uses a very small inductor.
C5, C6 are only 16 volt. The capacitance is much bigger.
Need to run some tests.
I tried a 3.5 volt RMS 50hz 0.1 ohm signal and could get 5V and 1A out.

 

[spec]

That circuit is totally unacceptable It does not comply with the customers specification. With a 100V output from the alternators the zener diodes, to give them their full name, will catch fire and explode (100v-24 = 76V. 76*1A= 76Watts). The circuit may work well and optimise the use of the energy from the cycle alternator, but that is only engineering.

For example, have you taken into account the impact on global warming? Will it provide refreshmant for the rider? What would happen if the rider was going down hill fast and streatched down and touched the terminals on the alternator! Have you taken the wider view and considered these issues- I think not!

In addition, on your circuit there is no MOSFET- you should add one to make the circuit work properly! If you don't know what MOSFETs are, I can help (image below)- I read all about them in a newspaper about 10 years ago. They are like tubes, which I used to design with, and are ideal for switching on and off power supplies, but you must not put any voltage on the gate electrode or it will punch across and generate a massive spark. Also, you must not touch MOSFETs without gloves on.

​

 

[spec]

Oh, by the way- nice work. I knew this project was in good hands

On a serious note, probably better to use more beefy Zeners than BZX84 (250mW) just to play safe or perhaps a BZX84 and a power tran to form a shunt- the cost increase would not be great.

 

[ACharnley]

I need to tell a story and see if you all agree.
Years ago I designed, something like this, for a large wind generator.
The current was (mega) and the spec said the voltage was 3,000 volts. Why 3kV?
In the case of a hurricane, AND the power grid went down so there was no place to dump the current, and the AND the 'lock the blades' function also failed, the generator would run-a-way and make 3kv.
So I made electronics that worked at high current under a short. AND In a over voltage condition it dumped heat into the steel tower. This kept the voltage to less than 1kV. So I did not design for 3kv because I made a load that kept the voltage down. My design had less parts, more efficient, low cost, less failure. The design was rejected because it did not meet specifications. The design worked well under a run-a-way condition.

SO: I think the alternator will not go to 100V with a 1 watt load. Probably not 50 volts. I can add a 2 watt resistor and small transistor etc that will not function below 50V but shifts in at 50V. We need to learn more about what 1 watt of load does at full speed!

That means the frequency is 9 or 18 times the RPMs. (?)
Is the dynamo connected to the hub so one turn of the tire=one turn of the alternator?

It's 18 poles and a bike wheel can rotate as little as say 3 times per second - 180RPM, although as stated I'm not expecting much watts until a higher speed.

The hub will give out 50V with little load at high speed. I know this by experience as it blew a LM2576 buck (45v rated) with a phone as a load. I don't know how much further the voltage will go, only that 50V is a minimum and 100V is what is talked about.

Unfortunately I don't have much kit to to more assertive measurements.

 

[ACharnley]

If it helps, Shimano make a regulator which dumps anything over 6V as heat. Consuming 1A, it gets quite hot during use, which shows there is some voltage over the 6V being produced.

It works the same as my twin darlington pair in the top circuit.

https://www.rosebikes.co.uk/article/shimano-sm-dh10-overvoltage-protection/aid:196629

I didn't realise the FET's have such a high switch on voltage. I take it the internal diodes are used until those switch on?

 

[spec]

If it helps, Shimano make a regulator which dumps anything over 6V as heat. Consuming 1A, it gets quite hot during use, which shows there is some voltage over the 6V being produced.

It works the same as my twin darlington pair in the top circuit.

https://www.rosebikes.co.uk/article/shimano-sm-dh10-overvoltage-protection/aid:196629

Brriliant only £3.58 too

I didn't realise the FET's have such a high switch on voltage. I take it the internal diodes are used until those switch on?

To fully turn big MOSFETs fully on you may hit them with 20V or so, just to make sure the gate capacitances are dicharged fast and the MOSFET turns on/off rapidly. A high gate drive voltage also ensures that the MOSFET remains in saturation as the inductor current ramps up, but smaller MOSFETs can have quite low threshold VGS (see attached image), the logic level MOSFET families for example. The chip that Ron has chosen will sort all this out. It will incorporate a relatively small pass element with correponding low drive requirements. It also has a voltage augmenting circuit.

Also, it is not mandatory to use a MOSFET pass element; you can use a normal medium power transitor, which will have a turn-on voltage around 500mV.

In any case you do not necessarily need to turn the MOSFET on hard with an inductive load, unlike a resitive or capacitive load. Because the voltage on an indictor = -L (dI/dT), the moment the MOSFET conducts any current, never mind how small the inductor voltage will saturate the transistor. As the current in the inductor ramps up though, you need to a correspondingly larger gate drive voltage to generate that current. Provided you turn the transistor off befor e it comes out of saturation all will be well. That's all part of the SMPS design.

When used as synchronous rectifiers, just like rectifier diodes, MOSFETS and bipolar transistors have to provide huge currents in a short time, so they especially need to be driven hard.

No, internal diodes will not do the job.

Where you only have say 5V supply rails for your circuit, you can generate an auxilary supply line of say 25V by a swinging capacitor techniqiue. Alternatively gate drive chips are around to do the job for you, as is the case with the LT1375.

Another approach is to use a gate drive transformer to increase the gate drive voltage- all pretty standard stuff.

PS: it is more difficult to design a switch mode power supply with a big and varying input to output voltage difference. That is why I showed an inverting type on the original outline schematic. They are particularly good in that sense.