| Author |
Message |
JiL0
Joined: 29 Jan 2004
Posts: 37
|
 15 Jul 2004 5:31 Transistor as Resistors |
|
|
|
Have anyone try using a transistor as a resistor before as a transistor can provide a large resistance with significantly smaller area? Is this common in anolog circuit design? Are they any drawbacks in doing so?
I've tried it on a simple 2 stage opamp to replace the bias current for the current mirror. Is it correct to do so?
|
|
| Back to top |
|
 |
guamak_menanak
Joined: 13 Apr 2004
Posts: 139
Location: malaysia
|
| 15 Jul 2004 8:47 Re: Transistor as Resistors |
|
|
|
A transistor acts like a resistor when Gate is connected to Source! This is happen because Vgs=0V!
Each component has their own pro and cons! Using active resistor can smaller your layout area, but it will effect on your f3dB and GBW as transistor has their own paracitic component! Having big (W/L) for the active resistor will effect your circuit performance!
I think there are not a big issue if you repalce the bias current with the active resistor! The problem occured when you like to define the node voltage at the drain!
|
|
| Back to top |
|
|
sunking
Joined: 25 May 2004
Posts: 914
Helped: 46
|
| 15 Jul 2004 10:20 Re: Transistor as Resistors |
|
|
|
it is usually that Transistor is as Resistors.
and more answer please upload you schematic
|
|
| Back to top |
|
|
maxwellequ
Joined: 27 Jun 2001
Posts: 185
Helped: 11
|
| 15 Jul 2004 12:13 Re: Transistor as Resistors |
|
|
|
To have a MOS transistor acting has a resistor you must ensure that it is on triode region (VDS<VGS-Vt). The value of the resistance is adjusted varying the VGS.
If you put VGS=0 the transistor is off (there is only a VERY SMALL current flowing – on the order of pA). It is not very useful as a resistor.
Resistors have parasitic capacitances between the poly and the substrate (although not always modeled). So, to implement a large resistance value it may be better to use a MOS, because you still get smaller parasitics.
The issue is usually the linearity of the MOS resistor – it is not that good.
|
|
| Back to top |
|
|
@lteravn
Joined: 22 Jun 2004
Posts: 3
|
| 16 Jul 2004 12:15 Transistor as Resistors |
|
|
|
I've used a JFET as variable transistor in controlling the amplitude of a sinewave generator!
It's great and in this application, I haven't found any drawback even it boost the frequency stablization of this circuit!
|
|
| Back to top |
|
|
@lteravn
Joined: 22 Jun 2004
Posts: 3
|
| 16 Jul 2004 12:15 Transistor as Resistors |
|
|
|
I've used a JFET as variable resistor in controlling the amplitude of a sinewave generator!
It's great and in this application, I haven't found any drawback even it boost the frequency stablization of this circuit!
|
|
| Back to top |
|
|
brain79
Joined: 20 Apr 2004
Posts: 57
Location: COLOMBIA
|
| 19 Jul 2004 18:05 Re: Transistor as Resistors |
|
|
|
| if you want transistors work like a resistance, bias it in triode region.
|
|
| Back to top |
|
|
djalli
Joined: 10 Nov 2001
Posts: 887
Helped: 15
Location: 1600 Pennsylvania Avenue, Washington DC 20500
|
| 19 Jul 2004 18:33 Re: Transistor as Resistors |
|
|
|
| JiL0 wrote: |
Have anyone try using a transistor as a resistor before as a transistor can provide a large resistance with significantly smaller area? Is this common in anolog circuit design? Are they any drawbacks in doing so?
I've tried it on a simple 2 stage opamp to replace the bias current for the current mirror. Is it correct to do so? |
Lets become frank JiL0. How large you want resistance JiL0? Give me a benchmark.
A 741 opamp has an input resistance of 2MegaOhms typically. I can go in market now and buy easily 20MegaOhm resistors and they are very small resistors. Put 2 in series and get 40MegaOhms. Last largest resistor I saw was available was 32MegaOhm thin-film resistor.
Why you want to complicate things?
|
|
| Back to top |
|
|
maxwellequ
Joined: 27 Jun 2001
Posts: 185
Helped: 11
|
| 19 Jul 2004 22:22 Re: Transistor as Resistors |
|
|
|
djalli,
Probably JiL0 is considering the design of an integrated circuit, where large resistors are really BIG.
|
|
| Back to top |
|
|
opamp741
Joined: 17 Jun 2004
Posts: 145
Helped: 10
Location: India
|
| 20 Jul 2004 7:07 Re: Transistor as Resistors |
|
|
|
MOS biased in Triode can be used as a resistor..
Only drawback is that resistance value changes with bias..
I m not sure how u hav used this resistor in ur opmap..but it doesnt sound good...
Regards
|
|
| Back to top |
|
|
djalli
Joined: 10 Nov 2001
Posts: 887
Helped: 15
Location: 1600 Pennsylvania Avenue, Washington DC 20500
|
| 20 Jul 2004 17:18 Re: Transistor as Resistors |
|
|
|
| maxwellequ wrote: |
djalli,
Probably JiL0 is considering the design of an integrated circuit, where large resistors are really BIG. |
I thought too but I changed my mind when he mentions he used "simple 2 stage opamp" to get a large resistance and I thought he possibly is looking for something with high resistance knowngly opamps have large input resistance. Then I said why dont buy 20megaohm resistances, they are in sale.
opamp741 above gave a good suggestion as using triode as resistor. There are endless possibilities in fact. You have to look for amplication where you want to apply it.
|
|
| Back to top |
|
|
JiL0
Joined: 29 Jan 2004
Posts: 37
|
| 21 Jul 2004 11:36 Transistor as Resistors |
|
|
|
| Sorry... wasnt around. Will be back with a diagram to make things clearer.
|
|
| Back to top |
|
|
JiL0
Joined: 29 Jan 2004
Posts: 37
|
| 21 Jul 2004 12:04 Re: Transistor as Resistors |
|
|
|
I'm refering to the circuit design of a two stage operational amplifier as shown in the diagram below.
When we design, we usually place a dc current source at the input of the current mirror. After everything is finalized, we determine the voltage at the point A and together with current known, we determine the resistor value to replace the dc current source.
So my question is,
a) is it possible to replace this resistor with a transistor where the drain and the gate are connected together with a resistor value of 1/gm?
b) or do we dont replace the current source in the first place and prepare an output pin to supply the necessary current externally from the chip?
c) is there a difference btw using a resistor and the transitor? ( i tried doing simulation using cadence software and both seeems to show the same result )
ps: sorry about the ugly pic. hope its clear
|
|
| Back to top |
|
|
brain79
Joined: 20 Apr 2004
Posts: 57
Location: COLOMBIA
|
| 21 Jul 2004 19:38 Re: Transistor as Resistors |
|
|
|
| you can biasing a current mirror with only add a diode-mode transistor, but, you have to add a start-up circuit because you can get a circuit with two answers, cero and your bias current. If you want to know more about self-biasing current mirror refer to Gray's book
|
|
| Back to top |
|
|
maxwellequ
Joined: 27 Jun 2001
Posts: 185
Helped: 11
|
| 21 Jul 2004 20:26 Re: Transistor as Resistors |
|
|
|
I though you were referring to a different thing ! Actually, if you put your transistor like that you have a diode and not a resistor (relation between voltage and current is quadratic and not linear).... but i now understood you...
First of all, the pmos transistor that it is connected to the point A (on the top left) must be connected as a diode, in order to mirror the bias current to the rest of the amplifier....
Regarding your question, you can put whatever you want (current source, resistor, diode) connected between node A and GND to set the bias current. The problem is the variations that you will have in the bias current, with the variations of the supply voltage, temperature and process.
If you use a resistor, then the bias current will be determined by the supply voltage and by the series of a diode connected transistor (the PMOS I mentioned before) and that resistor. For example, when the supply voltage or resistor value varies, your current will also vary. A similar situation is found if you use a NMOS diode instead of the resistor.
Note that it is not possible to create a large resistor with a MOS connected like this -> 1/gm is small for reasonable currents.
The only way to have a stable current is to put there a current source - the most simple is a transistor biased in the saturation region. Of course that, in this case, you must have a separated circuit which generates the bias current, that is then mirrored to your amplifier by this transistor (industrial designs use one circuit that generate a bias current, which is then mirrored to all the blocks of the circuit)
If your purpose is to make simple tests in the amplifier and you can add a pin just for this, the solution of controlling the bias current externally is probably the best one for you.
Hope this helps
|
|
| Back to top |
|
|
