KSC1845F strange curves

[superchromix]

Hi,

I recently tested several lots of transistors during my restoration of a Sansui AU555. I measured gain and Ic vs Vce using a PEAK atlas DCA Pro. I tested the following transistors: 2SC733BL (original from amp), 2SC733Y (NOS parts), 2N3416 (new), KSC945Y (new), KSA733G (new), 2SC458LB (original from amp), BC549 (new), KSC1845FTA (new), 2SC369BL (original from amp and NOS parts).

Along the way, I observed the expected gain curves for all of the transistors I tested, with the exception of one transistor model in particular: KSC1845FTA.

For example, here is a typcial set of curves for a matched pair of BC549:


And here is a pair of KSC1845FTA:


Note how the collector current rises slowly with Vce, at first, before saturating. This is not how the textbook says a transistor is supposed to work. The Ic vs Vce curve should rise steeply and then saturate at a constant (or slightly increasing) value, as shown for BC549 above.

Among the other transistors I tested, all of the curves appeared similar to those for BC549. The exception was KSC1845, which had this non-ideal shape.

Has anyone else seen this? Is this a defect, a bug, a feature?

cheers

 

[Watthour]

Is the test device/setup switching the transistor at a relatively high frequnecy?

 

[superchromix]

Is the test device/setup switching the transistor at a relatively high frequnecy?

I don't know exactly what you mean, but all devices were tested consistently, using the same tester. Testing was done using a PEAK atlas DCA Pro (DCA75).

https://www.peakelec.co.uk/acatalog/dca75-dca-pro-semiconductor-analyser.html

 

[Watthour]

I understand that, but their extremely limited "tech spec" listing does not reveal the test frequnecy at which gain is measured. The transistor's specifications have a frequeny contingent which affects operation.

For all we know, that could be tantamount to checking whether your cookies are baked by using a bathroom scale.

 

[superchromix]

I understand that, but their extremely limited "tech spec" listing does not reveal the test frequnecy at which gain is measured. The transistor's specifications have a frequeny contingent which affects operation.

For all we know, that could be tantamount to checking whether your cookies are baked by using a bathroom scale.

This is a fair point, but I could show you curves for each of the other transistor models I mentioned, and none have the unusual shape exhibited by the KSC1845...

 

[rcs16]

Might want to ask a more knowledgeable group over at DIYAudio or the manufacturer. Could compare against the original NEC parts if you can find any

 

[gslikker]

Probably testing higher voltage types of similar small signal transistors (2SC2240 etc) will give a much more similar result.

 

[Racingh11]

This is very interesting to me because I use those. Does it have to do with gain spec? These can be around 500, are any others that high? I have problems with some multimeters not accurately reading high gain transistors. They seem to like the 50 - 200 range better.

 

[Racingh11]

Also, what are the current ratings for the others? The 1845 is 50mA. Transistors don't have a flat response when pushed toward the upper limits of the current rating. At least from what I have seen. That curve you show matches the data sheet.

 

[sregor]

FWIW the curve matches what is in the data sheet. Why, I don't know, but the bottom line (IMHO any way) as it will make a poor switching transistor, and won't work well in circuits which cause the collector voltage to go to close to the emitter voltage. So, most circuits it won't matter. In a few it will.
I believe that this was the issue in this thread; https://audiokarma.org/forums/index...trange-clipping-problem.755929/#post-10286885, where the transistor was operating at a collector voltage of 4.8 volts and emitter voltage of 2. Switching the transistor to ksc1815 cured problem.

 

[Ylli]

Congratulations on that one sregor. One heck of a catch.

 

[superchromix]

This is very interesting to me because I use those. Does it have to do with gain spec? These can be around 500, are any others that high? I have problems with some multimeters not accurately reading high gain transistors. They seem to like the 50 - 200 range better.

Here are the specs for all of the transistors I mentioned. Other high gain transistors are also included in the sample, and do not exhibit curves that look like the KSC1845.



For example, here are the curves for two of the original 2SC369BL from my Sansui AU555 pre-amp stage, which have an hFE of ~400 (higher than the 1845s tested):



Curves measured for other devices in the list are posted in the following thread:

https://audiokarma.org/forums/index.php?threads/au-555-restoration.904525/page-2#post-13574841

 

[Racingh11]

FWIW the curve matches what is in the data sheet. Why, I don't know, but the bottom line (IMHO any way) as it will make a poor switching transistor, and won't work well in circuits which cause the collector voltage to go to close to the emitter voltage. So, most circuits it won't matter. In a few it will.
I believe that this was the issue in this thread; https://audiokarma.org/forums/index...trange-clipping-problem.755929/#post-10286885, where the transistor was operating at a collector voltage of 4.8 volts and emitter voltage of 2. Switching the transistor to ksc1815 cured problem.

I believe you are correct on that one. I have been using them for differential transistors where about half of the system voltage is across them at any time and they are operating at less than 2 mA. Fine for that, but I wouldn't want them driving anything. I think the curve they produce is directly related to high voltage combined with low current. If they were a 500mA or 625mA like many others, then the curve would be closer to what your seeing elsewhere. Even high powered output transistors start curving quite a bit at a little over half of the rated continuous current. I'm glad he started this thread as it is a friendly reminder to choose the correct transistor for the job.

 

[Racingh11]

It also appears to only be an issue if it's low collector-emmiter voltage and low base current. Since the 1845 datasheet lists Collector-Emmitter saturation voltage as typical .07V at Ic 10mA, Ib 1mA. Yet in the chart, it would have been over 2 volts at that Ic, but because it was at a low Ib. Highest on the chart is 16uA. This does make me think of another situation. If it is used in a circuit that does clip, and it is driven to the rail voltage; would it also round the clipping forms and produce a better, less harsh, form of clipping distortion?

 

[Pete B]

Quasi Saturation?
https://www.diyaudio.com/forums/sol...ignal-npn-quasi-saturation-measured-data.html

Many high voltage transistors (>50-100V) are far worse than low voltage transistors (<30 ish V)
with regard to quasi saturation. Doesn't matter much in a diff pair where the Vce is high.