Comparing Power BJTs (Sanken, ON Semi, ...) - diyAudio
Go Back   Home > Forums > Amplifiers > Solid State

Solid State Talk all about solid state amplification.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 22nd July 2010, 03:19 AM   #1
diyAudio Member
 
panson_hk's Avatar
 
Join Date: Aug 2005
Blog Entries: 10
Default Comparing Power BJTs (Sanken, ON Semi, ...)

I evaluate performance of my LME49811 amp with several power BJTs. Three pairs have been evaluated. They are Sanken 2SA1295/C3264, ON Semi MJL21194/93 and NJL3281D/1302D. The bias is 200 mA per device. There are two-pair employed. The supply is +/- 40 V (660VA transformer, 30000 uF per rail). The drivers in all cases are MJL150034/35.

Based on the results, 2SA1295/C3264 and NJL3281/1302 offer comparable performance. NJL outperforms Sanken pair in low-freq region. 21194/93 pair has the lowest fT. This is probably the reason behind its higher THD in high-freq region.

Here are THD vs freq plots for various loads.
  Reply With Quote
Old 22nd July 2010, 08:14 AM   #2
diyAudio Member
 
panson_hk's Avatar
 
Join Date: Aug 2005
Blog Entries: 10
Default THD vs level for 1 kHz tone

Here are plots for THD vs level for 1 kHz for various loads.

This is interesting to see that the Sanken pair is not as good as the 21194/93 pair in this test. The former has better hfe vs Ic than the later. The later has the highest clipping level for 2.7 Ohm load.
  Reply With Quote
Old 22nd July 2010, 08:37 AM   #3
diyAudio Member
 
panson_hk's Avatar
 
Join Date: Aug 2005
Blog Entries: 10
Default THD vs level for 10 kHz tone

Here a plots for THD vs level for 10 kHz.

21194/93 has the lowest clipping level for 4 Ohm load. This level is equal to that of 2.7 Ohm load. For the other two power BJTs, clipping level for 4 Ohm and 8 Ohm are identical.

THD for 21194/93 is flat from 5 V which indicates distortion limited performance. Sanken pair is more less the same, but with lower THD.

NJL performs the best in this test. The performance is still noise limited in high output level.
  Reply With Quote
Old 22nd July 2010, 08:49 AM   #4
diyAudio Member
 
panson_hk's Avatar
 
Join Date: Aug 2005
Blog Entries: 10
Default THD vs level for 20 kHz tone

Here are plots of THD vs level for 20 kHz.

Although 21194/93 is inferior to the others, the THD is still very low in common standard, < 0.01 % for 4 Ohm and 8 Ohm.

NJL and Sanken are below 0.01 % for all loads. NJL slightly outperforms Sanken in 4 Ohm and 8 Ohm cases.
  Reply With Quote
Old 22nd July 2010, 09:39 AM   #5
lineup is offline lineup  Sweden
diyAudio Member
 
lineup's Avatar
 
Join Date: Dec 2005
Location: the north
Good investigation. Thanks for publishing.

One parameter that is very valuable in a power amplifier system
is the gain of transistors used.
And when we have a look at those power transistors popular
we can see that they have a good high gain, hFE

An output stage with low current gain will put more load on the previous stage.
And so distortion increase.
We want light load.
This is also why triple darlington is used.
Such a stage is very easy to drive, due to the enormous current gain.
__________________
lineup
  Reply With Quote
Old 31st July 2010, 05:57 AM   #6
diyAudio Member
 
panson_hk's Avatar
 
Join Date: Aug 2005
Blog Entries: 10
Default Toshiba 2SC5200/A1943

Here are performance plots for Toshiba 2SC5200/A1943 power BJTs. They are employed in many Hi-End power amps, e.g. Luxman M-800A.

Up to this point, NJL3281D/1302D are the best pair among the tested devices.

Last edited by panson_hk; 31st July 2010 at 05:59 AM.
  Reply With Quote
Old 31st July 2010, 10:57 PM   #7
diyAudio Member
 
panson_hk's Avatar
 
Join Date: Aug 2005
Blog Entries: 10
Default Ic vs Vce curves

We all knew that a real BJT behaves more like an ideal BJT when handling lower current. Ic/Vce curves given in the data sheet are always not very useful for a particular amp operating condition. For instance, such curves of MJL21194 and NJL3281 cover Ic over 10 A. At such high Ic, current gain has dropped substantially, and not likely the normal listening output power.

I measured the Ic/Vce curves of the above BJTs for lower Ic. The DUT has a heatsink (65 x 40 x 20 mm) and cooling fan to avoid over temperature. The setting for curve tracer is written on the white label (mA/div, V/div). Note Ib was 5 mA for MJL21194 whereas Ib was 2 mA for other BJTs. What can see all BJTs have Ic nearly independent to Vce (ideal transistor Ic/Vce curves) when Ic is less than 2.5 A and Vce = 5 V. These are the Ic and Vce levels for a two-pair output stage driving a load consisting of 2 uF // 6 Ohm as shown in the simulation mode below where supply rails is +/- 40 V. The BJTs are nearly "ideal".

If the load resistor is reduced to 2 Ohm, the peak Ic will be about 7.2 A with Vce = 3.2 V. Hence, we also need a set of higher current Ic/Vce curves. They will be shown in the new post.
Attached Images
File Type: jpg 2SC3264.JPG (82.3 KB, 261 views)
File Type: jpg 2SC5200.JPG (105.0 KB, 156 views)
File Type: jpg MJL21194.JPG (101.1 KB, 124 views)
File Type: jpg NJL3281.JPG (104.2 KB, 121 views)
File Type: jpg sim model.JPG (150.3 KB, 282 views)

Last edited by panson_hk; 31st July 2010 at 11:00 PM.
  Reply With Quote
Old 31st July 2010, 10:59 PM   #8
diyAudio Member
 
panson_hk's Avatar
 
Join Date: Aug 2005
Blog Entries: 10
Default Ic vs Vce curves (cont.)

Here are higher current Ic/Vce curves. Although 21194 has a lower current gain, it seems to behave better in higher current compared to others. When Ic = 5 A and Vce = 5 V, it has a more flat Ic-Vce curve than that of the others. This may answer why output stage using 21194/93 results in a higher clipping level for heavy loads.
Attached Images
File Type: jpg 2SC3264 HC.JPG (109.4 KB, 265 views)
File Type: jpg 2SC5200 HC.JPG (110.6 KB, 85 views)
File Type: jpg NJL3281 HC.JPG (110.3 KB, 79 views)
File Type: jpg MJL21194 HC.JPG (111.8 KB, 87 views)

Last edited by panson_hk; 31st July 2010 at 11:13 PM.
  Reply With Quote
Old 31st July 2010, 11:23 PM   #9
diyAudio Member
 
Greg Erskine's Avatar
 
Join Date: Jan 2002
Location: Sydney/Australia
hi panson_hk,

Thanks. I am following what you are doing with interest.

regars
__________________
Greg Erskine
  Reply With Quote
Old 31st July 2010, 11:33 PM   #10
diyAudio Member
 
panson_hk's Avatar
 
Join Date: Aug 2005
Blog Entries: 10
Thanks Greg.

If we can parallel two more pairs, we should be able to get higher clipping levels. The new clipping levels for NJL, 2SC may be identical (slightly higher) than that of two-pair 21194/93. The new clipping level for 21194/93 may be two volts from the supply rail.

Last edited by panson_hk; 31st July 2010 at 11:36 PM.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
High linearty low-power BJTs and Class A headphone amp percy Headphone Systems 24 27th July 2009 05:00 PM
Looking for SPICE models of power BJTs Jorge Solid State 9 21st July 2005 01:13 PM
Toshiba JFETs BJTs Sanken BJTs Praise! Swap Meet 5 14th February 2004 06:45 AM
Toshiba power BJTs now at digikey mirlo Parts 8 28th February 2003 07:46 PM


New To Site? Need Help?

All times are GMT. The time now is 04:15 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2