Author Topic: Transistor Discussion (Read 13766 times)

jadaro2600 · « **on:** February 25, 2009, 12:39:06 AM »

For a greater understanding, I though that I would start a thread about transistors. Post any link to information you might have here about transistors ( just pile it on, even if it's a half baked idea ). Since one of the primary uses as of late has been for the Joule Thief Circuit, then it seems relevant to ask this / these question(s) here rather than muddy up an already lengthy thread:

And since I know no other location to post this...it lands here.

What happens when the transistor exceeds its switching frequency? - I assume it begins to heat up?

And, in the case of the Joule Thief Circuit, why would the current rather flow through the Light Emitting Diodes rather than through the Collector Emitter path of the transistor?

Any thoughts?

Low-Q · « **Reply #1 on:** February 25, 2009, 10:01:03 AM »

Bandwidth of a transistor is determined by the gain you want. This gain is determined by the input resistance and the negative feedback resistance. The transistors have something called HFE. HFE is the transistor gain without negative feedback - maximum gain. The higher gain is selected, the less the bandwidth. It happens when the switch frequency rate exceeds the transistor bandwidth, the gain goes down automatically. You have in transistors a C component that limits the bandwidth. C is a reactive component, so at high frequencies you'll have low impedance, thus more current flow, but at the same time lower voltage drop. So overall effect should be the same. Moreover, when this transistor to be connected in a circuit containing capacitive or inductive components, the transistor gain can lead to oscillation with the components it is controlling. Oscillation is caused by feedback from the circuit for a certain frequency is in phase with the signal out, and it the gain goes out of control. This may cause the transistor to overheat, and burns up. Therefore it is important to place small capacitors between the appropriate pins on the transistor so near it as possible. And for operation amplifiers, this is important. They tend to have a HFE for several millions.

(Google translate isn't THAT good

)

Br.

Vidar

gyulasun · « **Reply #2 on:** February 25, 2009, 12:22:10 PM »

Quote from: jadaro2600 on February 25, 2009, 12:39:06 AM

What happens when the transistor exceeds its switching frequency? - I assume it begins to heat up?

And, in the case of the Joule Thief Circuit, why would the current rather flow through the Light Emitting Diodes rather than through the Collector Emitter path of the transistor?

Any thoughts?

Yes, a transistor begins to heat up when driven by higher frequencies it is designed/manufactured for. The reason is the charge carriers inside are unable to follow the demanded quicker movements and their only possible "answer" is heat development. This happens even if you carefully operate your transistor well within its Safe Operational Area (SOA for short and you can find this in most of the data sheets). Of course this is a brief explanation but may reveal the main cause. HOWEVER, I do not think in case of the Joule thief circuit this is the main cause of heating because the switching frequencies involved here are relatively low for most of the transistor types used. I mean the so called transit frequency, f_T referred to in data sheets, it means the maximum frequency where the transistor gain reduces to unity. And if you consider a good rule of thumb of using 5 to 10 times higher transit frequency transistors for amplifying a signal of a given frequency then you realise that in case of JT here the max frequency involved is well under 100 kHz, 10 times of this gives 1 MHz and most of the transistors have got an f_T around or well above 1 MHz.

So I think the reason a transistor gets warm or develop heat in such JF circuits here is mainly due to power dissipation defined by its instantenous collector-emitter voltage and collector current (observing SOA conditions for that transistor type of course in any moment). You can reduce heat by using heat sinks on the body of a transistor and even in case of plastic and (not metal) 'casings' you may still use a gapped metal cylinder stuffed with heat conducting compound and clipped onto the plastic case to cool the body of a transistor.

Regarding the LED and the transistor current: do we speak about the same current that flows in both? Only their effective amplitude is what comparable. The current flowing through the collector-emitter path is determined mainly by the battery voltage and the coil's impedance (DC resistance and inductive reactance at the switching frequency). And the current of the LED is determined by the flyback pulse energy which is a function mainly of the self inductance of the coil, the current change through the coil and the rate of change of the current, ok? Although the flyback current (that comes from the collapsing magnetic field due to the sudden switch-off of the coil current) has got a much higher peak value than the collector current at switch-on, conventional physics explain it as the two currents' effective value can be the same only in an ideal loss-less component world, otherwise the effective collector current is always a bit higher than that of the flyback current.
This boils down to the following: if you want to reduce transistor dissipation to a minimum than you have to use types with minimum saturation collector-emitter voltage, once the collector current is a given quantity by the circuit design. There are many special switching type transistors manufactured with low saturation voltage, look for types with such feature, like in Kodak flash lights circuits for instance.

rgds, Gyula

jadaro2600 · « **Reply #3 on:** February 26, 2009, 02:04:47 AM »

Thank you all for your replies. Over the last few days I've been reading about transistors here and there. I'll be performing a few more experiments with the joule thief as I have it set up.

I'de like to add this link to this thread in any event - this person has quite a few videos about electron tubes and transistors with regard to radios: http://www.youtube.com/user/AllAmericanFiveRadio

electricme · « **Reply #4 on:** February 26, 2009, 02:16:09 AM »

@jadaro2600
@Low-Q
@gyulasun

First thanks to jadaro2600 for starting this thread.
I have thought the transisters being used on the Pirates JT thread needed further work or explination.

Secondly thanks to Low-Q and gyulasum for their explinations, which at my current level of understanding, I got drowned with the technical lession lol.
But my ears pricked up at the suggested usage of the Kodak transisters, now I think it makes sence to use them as they have been designed with high voltage switching.

So after sending this off, I will seek to tear apart a Kodak or Fuji or other camera that comes my way to experiment with.

But one last question, for all, which is the best device to use to drive a torid with a bit of current?
Should I use a MOSFET, NPN Transistor (2N3055) or a IGBT

I have a supply of transisters that drive the stepper motors out of Fisher and Paekel washing machines, these are found secured to the side of the water cooling heatsinks. Would these do? I just discovered that F&P have altered their heatsink design, no more water cooling, they are using G4RC10s transisters, hmmmmm.

General Jim

gyulasun · « **Reply #5 on:** February 26, 2009, 09:59:50 AM »

Quote from: electricme on February 26, 2009, 02:16:09 AM

....snip

But one last question, for all, which is the best device to use to drive a torid with a bit of current?
Should I use a MOSFET, NPN Transistor (2N3055) or a IGBT
I have a supply of transisters that drive the stepper motors out of Fisher and Paekel washing machines, these are found secured to the side of the water cooling heatsinks. Would these do? I just discovered that F&P have altered their heatsink design, no more water cooling, they are using G4RC10s transisters, hmmmmm.
....snip

Hi Jim,

If only I knew what you mean on a torid... Is it a mains toroid transformer or only a ferrite toroid core with some heavier wires?
Also, the frequency involved is also of importance. And your bit of current means 1 - 2 Ampers?

The G4RC10S transistor is an IGBT, see this data sheet if it makes sense:
http://www.irf.com/product-info/datasheets/data/irg4rc10s.pdf

The earlier transistors on the bigger heat sinks are probably also IGBTs or maybe power MOSFETs, if you write down their names (type numbers stamped on their casings), we may also find some data sheets on them.

rgds, Gyula

electricme · « **Reply #6 on:** February 26, 2009, 02:38:49 PM »

Hi Gyula,
I carn't remember, its 11.30pm night, an I need a bit of sleep lol, but I will try n think.

I am using a big torid, and also smaller ones which I have taken out of PC power supplies, they are yellow ones.

I havent got a clue about the frequencies that torids work best with.

If I could get 1 or 2 amps out of them, whacko, that would be great.

Its too late to go and look at the IGBT sheet, thats for tomorrow 1st thing.
I will also post the other transister numbers as well tomorrow.

Update on my JT project, I just wound 20 turns bifilar on each torid, they work OK.
Tomorrow I have to put 480 turns on each as the secondaries.
I want to have at least 8 torids, being energised by a single JT, which powers a 4017 IC, driven by a 555, modify the markspace ratio by using pots, then when the 9th LED on the 4017 lights, it sends a hi to another set of bc548's to dump the caps energy back into the 1.5v battery. Then the 10th LED hi I will rerout back to the reset pin on the 4017 IC so it will go and do its thing all over and over again.

see if I can get a bit of OU this way.

Gyula, thanks for the reply and your help so far.

Regards
General Jim

electricme · « **Reply #7 on:** February 27, 2009, 05:12:47 AM »

@ Gyula,
I just pulled apart a different F&P motor controller, it has a number of different transistors , 6 in fact, the IRF840 I think are the ones that drive the stater coils.

The others I will take a look for myself.

Thanks for your help and I will concentrate back on the JT forum.

Hooroo
General Jim

jadaro2600 · « **Reply #8 on:** February 27, 2009, 06:01:35 AM »

I've read this thread back and forth a few times now: Is it as a result of a path of lesser resistance that the LED in the joule thief lights up?

gyulasun · « **Reply #9 on:** February 27, 2009, 04:58:11 PM »

@General Jim

Here is a link to the IRF840, it is a power MOSFET with max 500V drain voltage and 8A drain current.
http://media.digikey.com/PDF/Data%20Sheets/ST%20Microelectronics%20PDFS/IRF840.pdf

Now I know what you meant by a torid...

Your earlier question was:
But one last question, for all, which is the best device to use to drive a torid with a bit of current?

It mainly depends on what circumstances you wish to use them, regarding their supply voltage and their collector or drain current. Because the devices' saturation voltage' V_CEsat given in their data sheets depend on these two data. Of course there are devices that are especially manufactured for just high speed switching with lower saturation voltages than other types for general purpose applications.
You can see a 2N3055 transistor collector-emitter saturation voltage in Fig. 4 and 5 (page 3) of its data sheet here http://www.qsl.net/on7pc/datasheet/transistors/2N3055.pdf You see the 2N3055 has a maximum of 0.1V saturation voltage from 0.1A to 1A collector current when you drive the base always with 1/10 of the actual collector current (I_C/I_B=10), from Fig. 5 ON voltages on the left. But in case of a 4A collector current @400mA base current, the V_CEsat increases to about 1.1V, this data is not from the Figures but from the tables ON characteristics, page 2.
In case of you IGBT device, G4RC10S, the saturation voltage is 1.1V at 2A collector current and 15V gate-emitter control voltage, data sheet does not give data on saturation voltage when collector current is lower.

So in case you have a battery (or supply) voltage in the order of 1-2V only it is obvious to choose a device with the lowest V_CEsat value because in that case the dissipation hence the loss on the device is at a minimal value too. Say you have a 1.2V battery and you wish to switch about 100mA current via a device through you coil, than the useful voltage available will be (1.2V - V_CEsat) for induction (here I did not consider the DC and AC losses of the coil). And if your device has ,say, 0.4V saturation voltage @100mA current, then it dissipates more than a device with 0.1V or even less saturation voltage, so efficiency suffers much in the first case.
Of course in case of higher supply voltages like 12V or 24V or higher, the saturation voltage become negligible compared to the supply voltage hence efficiency does not suffer so much from just this.

All in all, the 2N3055 is still a good choice both economically and saturation voltage points of view for you toroid driver. The IFR840 is even better, but it needs at least a 10V Gate-Source control voltage to bring its drain-source ON resistance under 1 Ohm (from its data sheet in the above link), in which case its dissipation would be just governed by its drain current (say the drain current is 100mA, then the voltage drop across the drain-source path would be under 0.1V , this would be its "saturation voltage").

Hopefully I shed some more light on your question...

Regards, Gyula

magpie · « **Reply #10 on:** February 27, 2009, 06:19:01 PM »

Would anyone happen to know of a compatible replacement for the BC547B transistor?
I am building some signal generators (the Velleman type that nievasoliveras posted) and don't have BC547B transistors.

Thanks.

qbjorn · « **Reply #11 on:** February 27, 2009, 09:25:48 PM »

Since this is a transistor discussion I'll take a shot:
Is there no one here with some more enlightened thoughts about William Fogal's transistor, http://www.eskimo.com/~ghawk/fogal_device/ ?
According to Bearden it should contain some spectacular scalar & potential properties (like infinite bandwidth and gain), but of course it was silenced - well, isn't that a surprising development ...
There not enough information in his patents (actually 2 different patents on the same device) to duplicate the device without some further knowledge, but Fogal was supposed to have come up
with a prototype device at his kitchen table which implies that anyone here - almost at least - should be able to do the same.
We just need to know the secret ingredient ...

electricme · « **Reply #12 on:** February 28, 2009, 12:26:36 AM »

@ Gyula,
Thanks for the handy link and the explinations to my questions, you are very knowledgeable with transisters.

This is a good thread here, and I can see it will have a lot of hits as soon as people know about it.

One last question, would this thread consider putting up a number of www addresses in the beginning here before it gets so big that one has to read every post to find info on a certain transister? It might save a lot of regular posts for simple questions, leaving you to answer only those which are considered the more unusual and need clarification.

eg Geramimium go to this http
Silicon go to this http
MOSFETs go to this http
IGBTs go to this http

20 odd years ago, I had access to the "Towers" data book which was considered at that time the best which had equilivents to each transister, is there a http address for this item?

Thanks again Gyula

General Jim

gyulasun · « **Reply #13 on:** February 28, 2009, 12:53:43 AM »

Quote from: jadaro2600 on February 27, 2009, 06:01:35 AM

I've read this thread back and forth a few times now: Is it as a result of a path of lesser resistance that the LED in the joule thief lights up?

Ok, let's discuss it once more.

Please study this link, http://www.ctglabs.com/soliton.htm it deals with testing Naudin earlier work on a (Caduceus) coil, my point is he switches a 12V battery across a coil (now it does not matter it is a Caduceus, it is still a coil of certain self inductance value and this is all that now counts in the explanation from your question point of view).

The scope shot shows the 5V input pulse from a generator (upper waveform, Naudin used a 10V pulse, Ctglabs used a 5V pulse) to operate Q1 as a switch and also shows the so called flyback pulse (lower waveform, Naudin calls it back EMF pulse) and you can see this pulse has about 56V peak amplitude, ok? It is measured with respect to the negative battery polarity. Vertical voltage scale is 10V/Divison for both waveforms. Flyback pulse is created whenever you interrupt the current flowing in a coil.

I edited the picture to show the pulse time durations when Q1 (IFR840 power MOSFET) switches ON and OFF. You can see that the 56V pulse appears across the drain and source points just after Q1 is switched OFF. Because the LED in a joule thief is also connected betwen these two points (i.e. between collector and emitter) it is obvious that this voltage is able to forward bias a LED (or several LEDs in series).

So whenever the transistor is just switched OFF in a joule thief circuit, the voltage level is just the flyback pulse between the collector and emitter where the LED is connected and current can flow through the LED because it "sees" a forward voltage polarity, the pulse's positive peak amplitude appears on the coil wire end that is connected to the collector, with respect to the emitter. The battery is a short circuit from AC point of view and I mention this because the pulse originally appears across the coil, and the upper wire connection of the coil that is connected to the positive battery pole can 'see' the emitter wire via the battery.

After the energy of this pulse is consumed by the LED, the LED becomes dark but human eye cannot detect it because it will be ON again very soon (a question of repetition frequency like we can see an incandescent lamp to light continuously when it has no current flow in it at the zero crossings of the 50 or 60Hz AC mains voltage). Notice the collector-emitter path has a high resistance value when the transistor is in a switched OFF state, hence no or very very little current can flow through it.

And whenever the transistor is switched ON, it means the collector point is pulled down to near zero voltage with respect to the emitter (just the saturation voltage could be measured by a scope) hence the LED cannot see any forward voltage to draw a forward current i.e to light up, ok? The transistor simply shunts the LED well below its forward voltage, current can flow through the transistor only.

Now the answer to your question: why would the current rather flow through the Light Emitting Diodes rather than through the Collector Emitter path of the transistor? is that in OFF state the transistor does NOT conduct current and the flyback pulse from the coil is able to forward bias the LED so that current can flow through the LED.

Hope you can see it all now?

rgds, Gyula

gyulasun · « **Reply #14 on:** February 28, 2009, 01:00:16 AM »

Quote from: magpie on February 27, 2009, 06:19:01 PM

Would anyone happen to know of a compatible replacement for the BC547B transistor?
I am building some signal generators (the Velleman type that nievasoliveras posted) and don't have BC547B transistors.

Thanks.

Hi, it is a pnp type, the 2N3906 is a possible substitute I think.

News:

User Menu

Author Topic: Transistor Discussion (Read 13766 times)

jadaro2600

Transistor Discussion

Low-Q

Re: Transistor Discussion

gyulasun

Re: Transistor Discussion

jadaro2600

Re: Transistor Discussion

electricme

Re: Transistor Discussion

gyulasun

Re: Transistor Discussion

electricme

Re: Transistor Discussion

electricme

Re: Transistor Discussion

jadaro2600

Re: Transistor Discussion

gyulasun

Re: Transistor Discussion

magpie

Re: Transistor Discussion

qbjorn

Re: Transistor Discussion

electricme

Re: Transistor Discussion

gyulasun

Re: Transistor Discussion

gyulasun

Re: Transistor Discussion