Hey
I was wondering about some of the graphs found on BJT data sheets, specifically the Ic-Vce and Hfe-Ic.
BJTs that simulate as being quite linear seem to have a Ic-Vce shape that looks like a pentode vacuum tube. The curves all rise up from the origin on top of each other on the left end, but break off at different levels and are horizontal lines going to the right.
There is also the Hfe-Ic curve. Even when the other curve looks wonky, this one might look very flat, as for the 2SC4883A in the data sheet attached. Transistors that give worse THD in the same test circuit have more of a curved line here, which suggests lower Early voltage according to experts I've read.
My assumption is that the Ic-Vce graph was an indicator of linearity, but some devices sim well and other don't. So it seems like the Hfe-Ic curve is actually more important. Can anyone verify if this ir correct?
2SC4793 always simulates with low-THD in any circuit I've tried it in, yet its Ic-Vce graph has the weird fanout appearance but its Hfe-Ic graph is fairly flat.
Are any of these curves (or some other one not mentioned here) a good indicator of linearity in a circuit?
I was wondering about some of the graphs found on BJT data sheets, specifically the Ic-Vce and Hfe-Ic.
BJTs that simulate as being quite linear seem to have a Ic-Vce shape that looks like a pentode vacuum tube. The curves all rise up from the origin on top of each other on the left end, but break off at different levels and are horizontal lines going to the right.
There is also the Hfe-Ic curve. Even when the other curve looks wonky, this one might look very flat, as for the 2SC4883A in the data sheet attached. Transistors that give worse THD in the same test circuit have more of a curved line here, which suggests lower Early voltage according to experts I've read.
My assumption is that the Ic-Vce graph was an indicator of linearity, but some devices sim well and other don't. So it seems like the Hfe-Ic curve is actually more important. Can anyone verify if this ir correct?
2SC4793 always simulates with low-THD in any circuit I've tried it in, yet its Ic-Vce graph has the weird fanout appearance but its Hfe-Ic graph is fairly flat.
Are any of these curves (or some other one not mentioned here) a good indicator of linearity in a circuit?
Last edited:
Depends entirely how you use it in the circuit.
All transistors' hFE goes to zero at zero current and at very high current (crystal gets congested). On older parts you might have only a 3mA-12mA zone where hFE varied less than 2:1. However even this made fine amplifiers if current-swing was limited, or if the complete amplifier had SO much current gain that hFE had little influence.
Likewise the pentode knee may be important if you swing ALL the way to zero (loudspeaker amps) and unimportant for small swings (generously fed preamps).
All transistors' hFE goes to zero at zero current and at very high current (crystal gets congested). On older parts you might have only a 3mA-12mA zone where hFE varied less than 2:1. However even this made fine amplifiers if current-swing was limited, or if the complete amplifier had SO much current gain that hFE had little influence.
Likewise the pentode knee may be important if you swing ALL the way to zero (loudspeaker amps) and unimportant for small swings (generously fed preamps).
Hi PRR
Thanks for the info.
Would you say then that the Hfe-Ic graph is a better indicator of general linearity? It seems that some amps have VAS CE stages that swing hardly any current and others swing a lot, offsetting whatever choice has been made in the current-booster after it - EF2 vs EF3, say. I would think in general that EF stages here would have wider current swings and maybe Hfe-Ic is a good indicator of THD?
I also wonder about the Ft-Ic curves: it seems the transistor has to be conducting above a certain current to have high-Ft. Is there an advantage for THD to make sure the Ft of the device AS USED is high?
Thanks for the info.
Would you say then that the Hfe-Ic graph is a better indicator of general linearity? It seems that some amps have VAS CE stages that swing hardly any current and others swing a lot, offsetting whatever choice has been made in the current-booster after it - EF2 vs EF3, say. I would think in general that EF stages here would have wider current swings and maybe Hfe-Ic is a good indicator of THD?
I also wonder about the Ft-Ic curves: it seems the transistor has to be conducting above a certain current to have high-Ft. Is there an advantage for THD to make sure the Ft of the device AS USED is high?
Just thinking aloud, since BJTs are current driven, current linearity should be the main parameter concerning transistor linearity and distortion.
If a BJT is used in a current-driven circuit then current linearity matters. Many BJTs are used in voltage-driven circuits where current linearity does not very much matter at all because the BJT is hugely (but predictably) nonlinear. There is a rule of thumb for BJTs: the peak signal level on Vbe in mV is equal to the second-order distortion in %.
Thanks. I see where it is coming from but I had never seen it stated so clearly.
Prof Cherry argues that all our devices are Charge Controlled. The BJT is "special" because it has a large and non-constant leakage on its charge. Much BJT design works to make it not-matter how much leakage a given BJT has.
I think I originally saw this in Foundations of Wireless, by Scroggie+Amos. Once you know this, it makes you smile when people try to make low distortion 'feedback free' amplifiers with BJTs. Every BJT is horribly nonlinear, but because it is reliably nonlinear we can use feedback to make very linear amplifiers. The only other way is to use distortion cancellation (e.g. balance) but that only deals with half the distortion.
Scroggie, figures, smart and pithy chap. Ordered a copy (rare in the US).
The thumb-rule is more than good enough for crystal-balling. The 8-digit idiot gets these numbers for a jellybean with zero NFB. Input in peak volts. Output is 5-harmonic THD.
0.03mV = 0.11% --suspect numeric noise
0.1mV = 0.1%
0.3mv = 0.23%
1 mV = 0.8%
3 mV = 2.3%
10 mV = 7.7% --not symmetric
30 mV = 21.6% --very bent
55 mV = 23% --gross
100mV = 29% --flat
While these numbers say THD%<mVpk, any external load will raise the THD.
Q4 et al is a bias supply to avoid diddling the bias manually. I biased DUT Q3 at 2/3rd rather than 1/2 because I knew it hardly matters for small output and at big output the wave would be big bottomed. I gimmicked the base-cap out and the numbers do not change.
The thumb-rule is more than good enough for crystal-balling. The 8-digit idiot gets these numbers for a jellybean with zero NFB. Input in peak volts. Output is 5-harmonic THD.
0.03mV = 0.11% --suspect numeric noise
0.1mV = 0.1%
0.3mv = 0.23%
1 mV = 0.8%
3 mV = 2.3%
10 mV = 7.7% --not symmetric
30 mV = 21.6% --very bent
55 mV = 23% --gross
100mV = 29% --flat
While these numbers say THD%<mVpk, any external load will raise the THD.
Q4 et al is a bias supply to avoid diddling the bias manually. I biased DUT Q3 at 2/3rd rather than 1/2 because I knew it hardly matters for small output and at big output the wave would be big bottomed. I gimmicked the base-cap out and the numbers do not change.
Attachments
Linearity in BJT's is not important as feedback in the amp adjusts for any none linearities.
I have even used IGBT's in amplifiers and the feedback fixes most ills.
- Status
- Not open for further replies.