Measuring base spreading resistance

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
At least there is a positive correlation! I was thinking of doing something along these lines too. I have a couple of questions. Vcesat is quoted for different combinations of collector and base currents, so which one do you pick from the datasheets? Also there are a number of devices in the Table 8.1 with noise around 0.2-0.3 nV, but you have only 3 dots in the figure below 0.5 nV/Hz. Maybe we need more statistics.
 
As DIY I would not go further than extracting Re from the log conformance and rbb from noise vs Ic.

I based my attempt on log conformance, and it sort of half-worked with older "low-noise" transistors, like the BC549-550 series, or the BC154.
It completely failed with more modern, higher current types, like the ZTX series.

Apparently, Rbb is only one of the many parameters that influence log conformance, and this is particularly the case with modern, advanced types.
 
At least there is a positive correlation! I was thinking of doing something along these lines too. I have a couple of questions. Vcesat is quoted for different combinations of collector and base currents, so which one do you pick from the datasheets? Also there are a number of devices in the Table 8.1 with noise around 0.2-0.3 nV, but you have only 3 dots in the figure below 0.5 nV/Hz. Maybe we need more statistics.

I didn't include the higher power transistors which have low noise but Vcesat ~1.0. 2SB1243Q, BU406 and 2SC3955. If you include these, the correlation becomes meaningless.

TAoE gives two columnar values for noise -- measured and optimized. Generally "measured" is @10mA.
 
Apparently, Rbb is only one of the many parameters that influence log conformance, and this is particularly the case with modern, advanced types.

You could search on things like Re compensation of log amps (very old school stuff). Parts like the MAT02/MAT03 were designed for both log conformance and low noise and I don't think their rbb is at an absolute minimum. These medium power devices are an example of re-purposing something away from its primary application.

If you can find it there was a paper on a device designed by Philips for Ortophon that was never made. The paper had a lot of geometric arguments on minimizing the parasitics that you won't find anywhere else.
 
Last edited:
I didn't include the higher power transistors which have low noise but Vcesat ~1.0. 2SB1243Q, BU406 and 2SC3955. If you include these, the correlation becomes meaningless.

TAoE gives two columnar values for noise -- measured and optimized. Generally "measured" is @10mA.

Vcesat on these is probably specified at 1-5A of Ic you would need to measure it at lower current for a proper comparison.
 
Most datasheets have plots of Vce(sat) as a function of collector current, so one can read off Vce(sat) at 10 mA. For example, for 2SC3955 it is about 120mV (while the maximum listed in the table of parameters is 1V).

The only other variability is ratio Ic/Ib used in the measurement, which can be in the range of 10 to 100, depending on the hfe of the transistor (typically several times smaller than hfe to drive the transistor into saturation).
 
Most datasheets have plots of Vce(sat) as a function of collector current, so one can read off Vce(sat) at 10 mA. For example, for 2SC3955 it is about 120mV (while the maximum listed in the table of parameters is 1V).

The only other variability is ratio Ic/Ib used in the measurement, which can be in the range of 10 to 100, depending on the hfe of the transistor (typically several times smaller than hfe to drive the transistor into saturation).

Where I could, I used the measured Ic parameter from table 8.1a and the 10:1 curve for Vce(sat) from datasheets. Quite a few of the devices have gone dodo even since the book was published. Don't throw a brick at me for using log-log chart:
 

Attachments

  • Noise_VceSAT2.png
    Noise_VceSAT2.png
    9.4 KB · Views: 198
This is cool, thanks for doing all the look-up!

Can you use the Excel option to label each point with the transistor number?
It would be good to understand what is the difference between transistors that have vce(sat) near 0.01V and very low noise vs. the ones that have vcesat of 0.005V but more noise.
 
Some of the data for Vse(sat) are very hard to read from the plots. For example, I downloaded the datasheet for ZTX851. Luckily its in a vector graphics format and I was able to make the lines of the plot thinner in Adobe Illustrator. Then one can see that for Ic=10 mA and Ic/Ib=10 the voltage Vce(sat) is only 1.6 mV.
 

Attachments

  • NPN_ZTX851_60V_5.0A_1.2W_Hfe75_TO-92a.pdf
    82.9 KB · Views: 75
It costs less than GBP 1,00 to purchase one. Now you can measure Vcesat at a variety of currents and a variety of temperatures.

Set the Vertical knob of your Tektronix 576 to "StepGen", the Horizontal knob to Collector and the Terminal Selector to "Emitter Grounded" and the vertical axis will display Ib, the horizontal axis Vce

(Unfortunately, the 576 doesn't give logarithmic results!)
 
Last edited:
This is cool, thanks for doing all the look-up!

Can you use the Excel option to label each point with the transistor number?
It would be good to understand what is the difference between transistors that have vce(sat) near 0.01V and very low noise vs. the ones that have vcesat of 0.005V but more noise.

I labeled the datapoints in Photoshop.

Anyone, feel free to provide better Vce(sat) numbers.
 

Attachments

  • VceSat_Noise3.png
    VceSat_Noise3.png
    77.6 KB · Views: 201
Thanks! I looked at more specific examples. For example 2sd2653 and 2sb1690 clearly have Vce(sat) of less than 10 mV, while 2sc3601 and 2sc3955 have Vce(sat) of about 0.1V. Yet the noise of the first pair is a factor of 2 higher than the noise of the second pair. So, while low Vce(sat) seems to contribute to giving a low-noise transistor, its not the only necessary ingredient.

Perhaps someone can stare at the examples of these datasheets and identify another reason why some transistors might be better than others.

I am going to get some PBSS4021nz transistors, which have Vce(sat) of less than 1 mV, the lowest I could find, and try to measure their noise. I haven't done such measurements before, so it may take me some time to setup the amplifier circuit. If anyone already has a setup to measure the noise, feel free to beat me to it.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.