Featured
Latest
Battery replacement...
 
Share:
Notifications
Clear all

Battery replacements for Valve Radios

55 Posts
3 Users
0 Reactions
39.4 K Views
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

Here is close up of the filtered output voltage (straight line) and the charging pulses at feedback point (Diode and R1) of the Dual LM393 or other comparator + Transistor/FET switcher.

Cyan trace is Gate drive. Rise time is the R C of the R5 1800 Ohms and the FET gate capacitance. For higher voltage FETs you need the LT1083 as a resistor pull up will be too slow. Fall time is the comparator open collector performance and gate capacitor discharge. Amplitude is the Battery Cells's supply voltage.

Forum 135
Click to see full size image

In theory it could work without the relaxation oscillator. In practice it will not self start!

There are so many cheap PSU ICs, that really this is only worth while if you have a dual comparator and no SMPSU IC in stock and want something quick. It's also maybe more flexible.

Max771 (obsolete)
Max1771 (replaces MAX771)
LM2623
TPS92010
UC3842
MC34060
MC34063  very suitable for high voltage circuits
Many more

Some are only low voltage in or low out
Some have switch connected to supply, so can only work as step down
Some have switch only to OV and can only step up.

 
Posted : 21/08/2011 5:54 pm
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

Also you will see loads of old LM723 and TL494  based Nixie and VFD drivers. Very obsolete chips.

 
Posted : 22/08/2011 2:04 am
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

Yes, the  NJM2360 is the 30c (US cent) "modern" replacement for classic MC34063.

I should probably get some, 10 off from Hong Kong / China likely about €7 in postage, well under customs limit.

Data sheet http://semicon.njr.co.jp/njr/hp/fileDow ... diaId=5894

It's actually almost identical to the LT1073 except the switch transistor only up to 40V. The best feature of LT1073 is the 50V rated transistor switch.

But LT1073 is a fixed 19kHz switching, which is grand for 950V MOSFET as they have 2200pF gate capacitance. But the NJM2360 uses an external timing capacitor. (100Hz to 100kHz).

Like LT1073 it's available SMT or 8 pin DIL

It's been highly recommended to me by an expert :)

So for Voltage Doubler to 75 to 90V, the LT1073 may be OK. But if using an external transistor or FET, then the NJM2360 is cheapest regulator there is.

EDIT
I bought 5 x NJM2360 just now for £6.99 ( €8.24) inc postage from Hong Kong. Likely have them on Friday or following week!

They are SO8 parts, not DIL, but easy to mount on veroboard by slitting the track to make two tracks either side of holes.

page 4 has the "common circuits". I'll draw up all the other topologies on the SMPSU article I plan for Techtir  

You can make all these with LT1073 or NJM2360
http://en.wikipedia.org/wiki/Sepic
http://en.wikipedia.org/wiki/%C4%86uk_converter
http://en.wikipedia.org/wiki/Buck_converter
http://en.wikipedia.org/wiki/Boost_converter
http://en.wikipedia.org/wiki/Buck-boost_converter
http://en.wikipedia.org/wiki/Split-pi

See also
http://en.wikipedia.org/wiki/DC-DC_converter

A Flyback converter or transformer is just a Boost Converter with the inductor having a secondary winding, or a subsidiary transformer in parallel with inductor, or capacitively coupled.

 
Posted : 22/08/2011 3:00 am
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

These MC34063 http://cgi.ebay.ie/5pcs-MC34063API-MC34063-34063-Switching-Regulators-/180652529327 are even cheaper.

or http://cgi.ebay.ie/20-SMPS-Controller-Standard-Universal-DC-DC-SOP-MC34063-/130537536978 (but it's SMT, not DIL, wrong photo!)

They could be re-marked NJM2360!

DIL version of MC34063 readily available (interchangeable with NJM2360)

one mad advert: (read what it says on chip!)

Forum 136

 
Posted : 22/08/2011 3:39 am
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

I built the voltage doubler version  using a "power" 100uH inductor by 14 turns of 0.56mm enamel wire on a miniature FT50-43 core. It was calculated to be 100uH and it measured about that.

Using 2.7k and 1M + 120K 1% feedback resistors I get 70V out at 6mA  (4 x 47K resistors), 74V with 3 and 80V with 2 x 47K parallel (3.4mA). Vin = 5V

It works down to 4V and up to 16V (though the LT1073 switch limits of current or volts may be exceeded.

I'll try the version with single rectifier and external IRF510 (cheap FET) next...

At 7.5V in and 82V out @ 3.5mA the supply is 82mA approx

82x 3.5 / 7.5 x 82 = 0.466, i.e.  approx 47% which is rubbish efficiency. But I might not have enough filtering on the input, so the DMM could be mis-reading as the current is pulsed.

 
Posted : 22/08/2011 4:44 pm
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

Hmmm....

Must check everything, with IRF510 I get 42V out, BUT the supply current is 1A to 2A!  😯  IRF510 thus is hot.
Something not right as the IRF510 should be switching fine at 8V supply. :(

Maybe I wired something wrong. It should work better than the doubler version.

 
Posted : 22/08/2011 5:11 pm
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

Back to the voltage doubler with an "off the shelf" 110uH or 120uH coil that looks like about 2A  or more

75V out to 4 x47k in parallel with each other and 47uF = approx 6.4mA

Pin = 7.56V @ 89mA = 673mW

Pout = 75V@ 6.4mA =  480mW

About 71% efficiency

5V in = 74.5V out

from 6V to 12V it's 74.9V out

The other expensive  meter agrees on output volts.

5.5V (flat battery 6 x cells) = 136mA, 74.7V out
6V in @ 128mA,  74.7 out
7V @ 105mA, 74.9 out
8V in I = 87mA, 74.9 out
9.5V  76mA,   74.9V out
10V 71ma, 74.9V out
11V 62ma (overcharging?) , 74.9 V out

Using 10x PP3 Zinc or Alkaline, or original B126  the voltages are exactly 55V (flat) to 100V fresh.

Peak voltage on regular switch is 42V (limit is 50)

This is with 2.7K from FB pin to ground and 820 + 150K (= 970K) as feedback divider.

With 1M instead of 970k, Vout = 77V and V switch = 44V peak

With 1M + 82k = 1082K (all 1%) the O/P is 78.5V

With 1M + 100K, O/P = 80V, switch = 45V to 46V

I think for safety I'll stick with 2.7K & 1M (1%) to set output, approx 78V

Approx 1mA no load

106mA connected to Radio. Now need to screen and filter it. (It works but Interference,  loads of long wires and plug in IC breadboard).

 
Posted : 22/08/2011 6:34 pm
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

Working Prototype!

Forum 137
Click to see full size image
The Startup transient response. 7.2V supply and 12K load  (just over 6.5mA @ 80V)

Forum 138
Click to see full size image
The Model for simulation. Not quite same as real schematic, which I'll publish as image and Eagle file.

Forum 139
Click to see full size image

1st version of Doubler on Breadboard. Sadly the toroid is saturating. Not quite right material. I'll source ready made suitable toroid. Meanwhile I have a stock coil that works (110uH to 120uH roughly).

Forum 140
Click to see full size image

Build it on veroboard to fit skinny metal box made of cut up coffee tin

Forum 141
Click to see full size image

Test with batteries (6 x C size cell, not fully charged.

Forum 142
Click to see full size image

Is the in-built transistor switch on the LT1073 running less than 50V with good margin?

Forum 143
Click to see full size image
10 volt per division. Yes it is at about 42V, limit is 50V

The output changes by less than 0.5V from 5V to 12V input change.
1mA on no load
106mA @ 8V supplying HT to Dx96 type 4 valve radio.

Maybe I'll build another neater one :)

A series 0.5 Ohms  (2 x 1 Ohm resistors across a 4 Ohm current loop Read relay) will sense the LT current and isolated switch automatically connect the 6 x C cell supply for HT inverter.

6 x C cells, current loop reed relay and screened. filtered electronics all fit in a "fake" B126 box.

It will have flying lead to the AD35 (which has 2 x D Alkaline) and then the two radio plugs, HT & LT plug into the B126 box.

 
Posted : 23/08/2011 1:19 am
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

Hi Ivan,

This looks very impressive and highly technical, but I'm afraid you've lost me  😆  

Are you designing and building a rechargeable battery pack?  😕

Best wishes,
Keith

 
Posted : 23/08/2011 8:46 am
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

Are you designing and building a rechargeable battery pack?  😕

Among other things. Yes, for HT. It works.

The Voltage doubler design is simple (most of the circuit is filtering!) but only  useful up to 80V. Which is fine actually for 90V Radio.

It's easily adapted by changing one resistor for 22.5, 45V, 67.5V Battery HT.

For 120V upward a different technique is needed. I'm working on that too.

So this thread has various thoughts and discussion on

  • connectors needed (I have good method now to make sockets out of cut up coffee tin and blank single sided paxolin PCB)[/*:m:ciu01yn1]
  • Size of packs and images to print to make boxes[/*:m:ciu01yn1]
  • Different batteries to use to stuff the boxes, disposable and rechargeable[/*:m:ciu01yn1]
  • How to automatically turn HT pack electronics on/off[/*:m:ciu01yn1]
  • Different circuits, ICs and parts to make Step up inverters[/*:m:ciu01yn1][/list:u:ciu01yn1]
  • The "Batterymaker" makes very nice battery packs, but most DC to DC stepup inverters sold for VFD, Valve battery portables and Nixies seem quite poor value, or poor designs or obsolete.

    The LT1073 is a very old IC, but still in production (I got free sample) and over all the others I looked at is only one to manage 50V directly. The next best is 40. Using a 40V rated IC, you could at best reliably have about 60 to 62V HT using a doubler. For higher you need external FET or BJT (High voltage power Transistor).

     
Posted : 23/08/2011 9:45 am
crustytv
(@crustytv)
Posts: 12446
Vrat Founder Admin
 

That last scope trace looks remarkably like a TV's Line output transformer ring trace.

This is an altogether fascinating thread, thanks for taking the time to post it. As I've said before over the shoulder logs of this kind are invaluable, I encourage everyone if your restoring a radio, TV, test equipment or experimenting, this type of post is so useful to those that follow in your footsteps.

Keep up the great work Mike

Chris

CrustyTV Television Shop: Take a virtual tour
Crusty's TV/VCR Collection: View my collection
Crustys Youtube Channel: My stuff
Crusty's 70s Lounge: Take a peek

 
Posted : 23/08/2011 9:50 am
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

That would be because an LOPT is working by same method to create a "boost" voltage that is then stepped up by secondary to EHT :)

Actually if you have say 20V to 45V step up inverter for a "main supply" and need 200V to 450V you just add a secondary winding on the inductor.

The disadvantage for use is that a Flyback transformer is not off the shelf. A basic inductor is. So we have to either use doubler or tripler to get above 45V, or external high voltage transistor.

The problem is that the higher the step up voltage, the shorter and higher current the on pulse.  Losses get higher.

The sharp eyed may have noticed a cheap small 1N4148 has crept in as -ve voltage clamp (only occurs at start up time). Why big fat 1A Ultra fast silicon rectifiers rather than 1N4148 (the radio is only < 7mA)?

Of the top of my head ...
Well, the 45V pulse can be only 2% of time. That's then a current of 350mA for 7mA average. But  total pulse current is 700mA as there is two 45V supplies in series. You want low loss in the diode.  

Good luck trying to find a 120V Schottky rectifier too!

If the on time of switch is only 5%, then the peak on current is 5A for 106mA. No wonder my home made Toriod was saturating with more than 2mA load current.

I think though it's not as bad as that. maybe 900A peak when average is about 100mA

The 68 Ohms on LT1073 limit pin ensures internal transistor switch doesn't see too much load current, somehow.

 
Posted : 23/08/2011 11:13 am
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

Thanks for the explanation Mike, clever stuff.

I have scanned some more pages from the GEC catalogue. It's dated 1958. Hope someone finds them useful.

Keith
Forum 144
Click to see full size image

Forum 145
Click to see full size image

Forum 146
Click to see full size image

Forum 147
Click to see full size image

Forum 148
Click to see full size image

Forum 149
Click to see full size image

 
Posted : 23/08/2011 11:44 am
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

Thanks.

I found the earlier ones good.

I checked the coil and Switch transistor current (wrong feedback of 970K instead of 1M, so currents will be a about  3% higher.

Forum 150
Click to see full size image

When the capacitors are all charged fully, the PSU cycles on and off. This is more optimal than varying frequency or duty cycle.
Green = Unfiltered Voltage to feedback
Thin line = fully filtered output
Cyan = Coil current

Zoom in about 220ms when initial capacitor charging is done:
Forum 151
Click to see full size image

So under 1A peak which is good :)   Average is about 106mA @ 8V. Or 220mA @ 4V (below 5.5V is not a good idea).

Unlike direct battery supply there is no warning that battery is getting low. The performance will fall off a cliff when batteries go flat. Really there should be a beeper that sounds periodically when below a certain volts with more rapid beeps below 5.6V, then disconnection at 5.5V

 
Posted : 23/08/2011 11:57 am
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

The Boost or Flyback Inverters are electrically very noisy.

So I had a think and decided to make a Ćuk inverter for comparison. The "standard" circuit has no isolation so has problem that the O/P is negative. You could  simply connect the battery + to 0V, but that creates problems too.

I liked idea of a small transformer for isolation and then you could exceed the 78V if desired.

I did have one briefly working, and blew up the chip! A 2nd one only worked well below 0.5mA load.

So I researched a bit more and did some more detailed simulation (it's only simulating about 25us to 50us per second and you need to simulate between 25 to 250ms to see what is happening. :(  So that's 1000 to 5000 second of our time!

Anyway, nearly there. Because the currents are longer pulses you can use 2 x 1N4148 diodes (only 1 has the HT volts) as the current ratings not exceeded instead of 1 x 1N4148 and 3 x MUR120.

Also the switching ripple before filtering is about 1/10th or less.

 
Posted : 24/08/2011 11:15 pm
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

I think I've "cracked" the transformer based Ćuk inverter.
Forum 152
Click to see full size image

L1 & L4 are "coupled" inductors. It works if they are not a transformer, but reduces ripple a lot if they are. That's a standard magic of this type of converter.

The  Ćuk inverter normally doesn't have L3, L5 and C4. D1 normally would be from C2 to ground giving a negative output. See LM2611 datasheet.
L3 & L5 are a transformer. Because it's DC isolation you can re-arrange the parts and have a + output. For high voltage, High power (like a 100W valve amp 500V HT off a car battery) D1 is replaced by a 10A 950V Mosfet that is switched on when D1 would have been conducting (so any inherent Source Drain diode never conducts).

V1 is the battery pack.

the 12K & 8uF C5 simulates the radio.

switching noise is very low.  It's essentially symmetrical, so in theory when running of a 90V mains adaptor it could be switched to be a charger.

To Charge instead of supply:
The "FB" pin sense battery current and SW1 + SW2 pins control a High voltage FET switch across D1.

The Transient response at power on:
Forum 153
Click to see full size image

Yellow is output and blue is switch transistor.
Red is Feedback sense point.

Notice absence of HF noise, just LF ripple. We can increase C8 if needed.
We can also go to maybe 110V without overloading the Switch Transistor.

 
Posted : 25/08/2011 2:10 am
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

Hmm. The  inductance values vs believable materials and flux seemed not achievable.

Revised version that *might* work with FT50-43 ferrite toroid core.

1:4 Step up transformer
L5 = 84T, 2.65mH, 0.422 Ohms using 0.26mm wire, fills core with one layer. 128cm + 10cm to wind it.
Then wind L3 on top using 21T, 165uH, 0.023 Ohms using 0.56mm wire. Space evenly. (up 0.9mm wire will fit too in theory)

Wind both coils same direction, then you can easily decide o is the same on each.

Repeat for the current limit Inductors L1 & L4 that are wound like a transformer.

C2 and C4 do the actual power switching so their capacitance ratio must be equal to TURNS ratio of transformers/coil, not the inductances (the Inductance ratio is approximately the square of turns ratio).

I'll test the "real" circuit soon!

 
Forum 154
Click to see full size image

Note the typical Battery Radio Mains Adaptor uses 2k7 Ohms in series between two 32 uF electrolytics.
Green is feedback point, Red is Switch transistor Collector inside IC, only 30V, so the cheap MC34060 (obsolete), MC34063 (current) or NJM2360  can be used in same circuit.

Forum 155
Click to see full size image

Forum 156
Click to see full size image

zooming in on the ripple looks like 235kHz (though the switcher is 19kHz!)

The output is clean.

I figured how to drive a battery low beeper and cut HT to save rechargeable cells at the 0.9V to 0.95V per cell point.

 
Posted : 25/08/2011 1:24 pm
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

Another value of lowering the switching voltage is that it reduces the flux and risk of core saturation

At 100kHz, FT50-43 and 165uH (21 turns)

RMS Volts  and Flux
5  404
10 807
15 1211
30 2422
45 3633

Moving to 84 Turns (2650uH)
RMS Volts and  Flux
10 201
15 302
45 907
120 2417

Unsurprisingly the turns you need are proportional to the voltage.

2650uH (84T)
At 24V and 100kHz, flux is 483 and max flux 500

At 1MHz, the max flux drops to 150, but actual flux is 48, so max voltage rises from 24V to 74V, approx. Basically Square root of frequency, or the other way, if you have 1/10th frequency you need about 3.2x as many turns.
i.e. 200KHz transformer needs 84T for 34V RMS for type 43 ferrite.
so 20kHz needs about 250T. You'd need an FT114 core instead of FT50 to fit the wire. such a core / turns is good for 270V at 200kHz and might manage 84VRMS at 20KHz.

Unfortunately calculating the core flux with the Cuk waveform is not something I have done. I know the FT50-43 wound only about 8 turns on Boost inverter (27uH) could manage 1mA at 70V, with 7V supply. Increasing load or supply reduced the O/P and rapidly increased current showing core was saturating.

 
Posted : 25/08/2011 2:32 pm
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

Actually a FT50-78 is a better PSU choice (up to 5W maybe), larger core for higher power.

An FT114-43 might work if a FT50-43 is marginal.

I have two coils/transformers made now with FT50-43 test later.

the
FT = Ferrite Toroid
50 = size
-43 = material, type of Ferrite.

FT50-43 are widely available and cheap. You need a proper distributor of Amidon or Fare-rite to get the 78 material cores.

 
Posted : 25/08/2011 5:48 pm
Anonymous
(@anonymous)
Posts: 16851
Group Deactivated Account
Topic starter
 

Gratifyingly the Ćuk converter burst to life with 83V when I turned on the Bench PSU (7.5V). I like when things work 1st time :)

Except of course as expected, it's not quite right. The FT50-43 cores I think are close to saturation, but I'll increase C2 (currently 0.66uF) and C4 (currently 0.2uF), the charge pumps, just in case.

Forum 157

The waveform:

Forum 158

Notice hardly any ringing (more efficiency, less interference) compared to Boost or Flyback converter. With the Boost + doubler the IC switch was running at 43V for 76V out and 3x 1A ultrafast diodes needed as well as 1N4148 switch clamp. I'd like to know why a second switch "on" and O/P pulse though.

Repeat of schematic in Spice earlier:
Forum 154
Click to see full size image

This uses just two 1N4148, one is clamp and the other is a commutator switch rather than a "real" rectifier as C2 is a charge pump isolated to by L3&L5 (K2 transformer) become  C2 & C4.

The IC switch is running just at 30V. The flat part of switch trace is PSU volts of 7.5V (i.e. Battery) and zeros before the 30V spikes is switch on time.

I'll fiddle with it a bit. But I think I need type 78 ferrite core or something.  It will be nice if I can improve it for FT50-43 as they are so cheap and common.

Zero Load = 84V
1.8mA = 81.7V
4.4mA =  73.2V

So not as good as the Boost Inverter with an "Off the shelf" 110uH coil for load,  but much less filtering.

However no external transistors and the lower voltage rated  cheap MC34060 (obsolete), MC34063 (current) or NJM2360 will work in circuit (may need to change Feedback resistor value R2, I haven't checked.)

 
Posted : 25/08/2011 9:48 pm
Page 2 / 3
Share: