Ah ha!
I was running only on 0.22, not 0.66uF, the coil and one capacitor share a different column to the other two! You can see it in the photo!
Putting all three together much improves it. I'll try the full approx 2uF of spice model next rather than just 0.66uF (C4 minimum = C2 / 4 and C2 minimum, the charge to transfer.
Analysing the spice model and looking at Scope I'm convinced the LT1073 isn't switching in correct way for ideal Cuk converter.
So I knocked up a spice model of a comparator wired as relaxation oscillator and drive the + feed back reference from an integrator adjusted by a comparator comparing a reference 4.7V with the HT.
The osc drives a fet switch. With correct design and values it should not "burst" on/off creating ripple, but reach a steady state. The L1 & L4 should not be energy stores, but lossless constant current sources.
The advantage of a separate FET is that you can just use 1:1 pulse transformers, baluns or common mode filters "off the shelf" as it can run to the HT voltage.
Click to see full size image
Almost no output filter ...
"Sense" is connected to o/p coil of the converter. No other filters.
Click to see full size image
Blue is unfiltered "Sense" point. the other is O/P on load.
Click to see full size image
Zoom in detail of raw unfiltered HT at feedback sense point
Click to see full size image
Zoom in on filtered O/P. Just drift. No visible RF or ripple.
The FET switch peaks at about 120V for the 104V out. 1N4148 no good at those volts either. MUR460 or similar.
Here is current model.
An LM393 should work.
Click to see full size image
L1 & L4 are not "boost"/Flyback coils, but current sources to limit rate of change of current to switching capacitors C2 and C4. They are wound on same core like a common mode choke or balun, and 1:1 in this design, so available as an "off the shelf part"
L3 & L5 is a pulse transformer to allow the polarity to be positive by splitting C2 into C2 + C4
U2 could be powered from output via Zener and resistor driving an opto-isolator for fully isolated PSU.
Note wires + crossing are never joined, the JPEG compression only make them looked joined.
C12 with R9 to R12 form a "relaxation oscillator" U1
R7 is because comparators are "open collector". In theory a high speed op-amp able to source and sink the current at 40kHz on the FET (M1) gate capacitance would work better.
There is a guy in UK selling 100 x CR2032 cell packs for a about £5. But he doesn't ship outside UK.
So I ordered 200 cells for €15 inc postage from HK/China. That's less than the €20 customs duty/VAT limit and 7.5c a cell inc postage.
I'll test capacity, but they should be by volume about x3 better than Alkaline PP3 and x15 better than Carbon Zinc.
10 x Carbon Zinc PP3 about 45mAH and ??? cost?
10 x Alkaline in Aldi about €17.50 and 200mAH
Both are about 96V fresh dropping to 54 when "flat"
30 x Lithum cells = About 93 to 94V on load dropping 82V when flat, 220mAH, so superior. But you need 3 or 4 stack to have decent current capacity.
A B126 has space for 150 coin cells with 18mm depth of box still empty
top view of coin stacks in box:
_____
|OOO|
|OOO|
That's 5 x 30 = 90V nominal @ 1000mA nominal for (15 / 200) * 150 = €11.25
Intermittent shelf life over 10 years! (NiMH self discharge in 10 days to 2 months, eneloop and NiCD are better).
At 6.5mA that's 153hrs, or about 7c an hour. Or at 4 hrs a day over a month.
If you stack 2 sets each side of connector you can fit another 12.
Total = 162
If those are split to 6 stacks, its 27 per set = 81V nominal, but end point is 74V! Still within operation range of radio. This would increase capacity to 1200mAH to 1680mA depending on cell quality, Resistive losses are less by 1/6th too. Duration could be 185 to 250 hours for less than €15.
Side view
28 25 28 back
28 25 28 front
Total 162
I will have my rechargeable pack. But it's about 23 hours use between charges. It will be nice to have a long life pack that doesn't go flat in storage!
Be interesting to see what quality they are!
Just for information, my ten PP3 HT battery cost £5.
- Joe
Just for information, my ten PP3 HT battery cost £5.
- Joe
Definitely Alkaline or generic Zinc Carbon Joe?
I can get at that price online (Zinc Carbon), but I've not seen genuine Alkaline at that price. Was it online or a local "Pound shop"?
Cheapest local shop Zinc-Carbon are €1 each and Alkaline about €1.75 each (thus better value by far)
Cost of 162 CR2032 coins is 15/200 x 162 = €12.50 (About £10?)
Capacity worst maybe 1200mAH (200mAH x 6), Panasonic claim 280mAH, others claim 220mAH.
Alkaline PP3 is at best 200mAH at low current. Zinc-Carbon between 40mAH and 50mAH at best.
so if you battery are Alkaline per hr listening the coin cells are a bit more than 1/3rd price and if they are Zinc Carbon, then per hour listening the about 1/15th the cost.
But mainly I liked the idea of trying something different as "everyone" stuffs with PP3s.
I'm not sure but would imagine that they are Zn/C. It says on them "SONY DRY BATTERY 6F22/S-006(P) 9V ULTRA SUPER". They did indeed come from PoundLand.
When it is flat I intend doing something more sophisticated, if only for the fun of it. This was the easiest and cheapest solution to make sure that the radio worked.
How about a re-stuffed PP9? It's just this low maximum current drain that puts me off somewhat. I take it that if you wanted 20ma you'd need 20 sticks of three.
Interesting as ever, even if some of the technicalities are way over my head. I'm looking forward to the final design of you step up PSU and shopping list of bits.
Thanks, Joe. Hadn't thought of PP6, but actually makes more sense than AA rechargeables that go flat even when not using them.
You can draw up to 5mA per "stick" of cells. But at decreased life time. The 1mA is a good target/
PP9s are available, just expensive and scarce.
The Fidelity actually takes 15 to 25mA depending on volume.
The PP6 is similar to B126 but shorter. You can fix six stacks of 20 coins = 120 cells easily in a PP6.
20 sets of 3 = 20mA to 40mA @ 9V nominal and 600mAH
40 sets of 3 = 40mA to 80mA @ 9V nominal and 1200mAH
So you could 1/2 fill a PP9 (60 coins), and cost is about €4.50 in postage.
Completely (120 coins) filling it is about €9 worth of coins and since nearly full, about x15 life of a real Zinc Carbon PP6. Probably about a year's usage?
Quality is more consistent as end point is 8.25V instead of 5.4V (Alkaline or Zinc Carbon or Rechargeable 9V battery end point).
Sensing LT current to switch on/off HT DC-DC converter electronics.
Using 5 layers of 0.26mm enamel copper wire on a sub-miniature reed switch is 1.6 Ohms and current sensitivity of 83mA. Using 4 layers isn't reliable at 120mA
At 125mA the voltage drop is 0.2V.
For a higher current LT radio 3 layers will be enough.
I'll try winding the 5 layers a bit shorter and tighter.
I wound the coil just the length of the reed switch on paper on a screwdriver blade more neatly with 5 & 1/2 layers, 0.32mm enamel copper.
Pull in at 109mA
Drop out at 61mA
Probably 500mA will do no harm
So even a capacitor after the coil will add enough of a pulse that a steady current of 100mA works.
At 125mA the voltage drop is 82mV. Measuring resistances less than a couple of ohms accurately is hopeless on the DMM.
It's just over 2m of wire I think, 2m of 0.32m is 0.435 Ohms
The voltage & current suggest 0.656 Ohms
The DMM says 0.6 or 0.7 Ohm if you hold the wires tight and wait a while
So I think a bare reed is a viable alternative to a current loop relay to detect LT 1.5 "A" battery draw and turn on the the electronics for the "B" HT supply inverted from 6V to 8V (6 x C cells)
A quick ebay UK search suggests
£2 for 2 inc postage
http://cgi.ebay.ie/Reed-Proximity-Switc ... 0335454444
£2.50 for 3 inc postage
http://cgi.ebay.ie/Reed-Switch-Tiny-Sma ... 0659641512
Or 10 off for $6 inc postage from China
http://cgi.ebay.ie/Reed-Switch-Glass-N- ... 0365859143
6 for $4 inc postage in China
http://cgi.ebay.ie/Reed-Switch-GlassN-O ... 0844586377
Most Reed relays are voltage coils, 5V to 48V. Current loop 4 Ohm parts do exist but they are very expensive. It doesn't take long to wind your own current actuated reed. The 0.32mm wire ( is a lot less fiddly than the 0.26mm and gives I think low enough resistance.
Well I stuck $1 bid on 10 reed switches and got them. €4.27 inc postage from China
There are many items on ebay like that. I bought 20 12kV fully encapsulated diodes for $10 including postage - much cheaper than EY51s! It's well worth scanning the electronic components sections.
200 off CR2032 coin cells came today. I'll test the real mAH later, should be between 200mAH and 280mAH.
6 tubes each of 26 cells = about 86V HT. But end point of Zinc Carbon B126 is 0.9V x 60 = 54V approx.
End Point of Lithium Manganese DiOxide is 2.75v, so 2.75 x 26 = 71.5V
1N4148 x 6 used as combiner in -ve end, so volts is from 85V approx to 71V
I'd previously made a LT pack, AD35 with velcro fastened lid and 2 x Alkaline D cell paralleled and "coffee tin" connections
Click to see full size image
€14 inc postage for 200, a 156 used.
Each tube of cells checked and with +/- 0.3V of average.
Sadly the coins taped into trays with parcel width clear tape and "gum" hard to clean off cells.
Click to see full size image
Diode combiner & Top (negative) terminal
Click to see full size image
HT socket made of coffee tin and blank PCB
Click to see full size image
Click to see full size image
Click to see full size image
Click to see full size image
Bottom of pack, coffee tine positive plate.
Tubes are thin card.
casing is marge tub.
Thinking about this:
The tubes are awkward to fill
The top and bottom connections awkward to attach securely to tubes.
There is some space.
Coins are 20mm diameter, so 40 x 60
The base of B126 is 45mm x 62mm approx (a bit larger perhaps).
So maybe the solution is clamp 3 pieces of wood 85mm long
middle part 22mm thick x 62mm wide and two outer parts 11.5mm and drill 6 x 20mm holes. Then top and bottom plates can screw on and the sides can lift off to give "trays" to fill with the coins with end plates still attached.
?
Anyhow it works and should be about x5 life of 10 x Alkaline PP3 or x20 life of 10 x Zinc Carbon PP3.
(6 x 200mAH stacks vs 1x 200mAH or 1 x 50mAH)
My miniature cheap reed switches came.
One was miniature, the other 9 positively minuscule
Bottom is original 5 & 1/2 layer coil with original reed I had lying about.
Middle reed switch is the odd one out of the 10 ordered.
Top reed is one of 9.
Top is the smallest reed in a 4 layer 0.26mm wire coil.
Pulls in at 108mA.
About 95mV series voltage loss @ 125mA, which is a little high.
Drops out about 60mA
I'll try 5 layers of 0.32mm wire next. That should be less than 80mV drop
For 250mA Dx9n series battery radio it's likely 3 layers of 0.32mm are fine.
Conclusion
very cheap miniature reed switch with a coil that takes less than 15min to make is effective LT current sense to operate an inverter for HT. Easily adjusted for type of radio.
