Hiraga 20W class A

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Nice work, sounds like you are getting close

Thanks for the encouragement.
I raised the 0.33 ohm resistors to 0.5 and the result is aprox the same 1.01V across them =>2.02A bias current from 0.734V across the 0.33Ohm (2.22A).
Looks good but i increased the 0.33 ohm for this. I'm at aprox the same value as the 20w Super class A wich had 0.47 ohm resistors.. :(
 
Last edited:
Daniel, Gaborbella, Hugh, 6L6 look what i did now :). Look at the picture bellow with the results.
I cannot decrease the bias with other method than by lowering the 1k8 (and 1k5) resitors and injecting some more negative and positive currents into the output transistors bases ( i see that if you increase the two 1k8 resistors nothing dramatic happends on idle, but if you decrease them your bias goes down).
I have a problem with one of the two 1k8 (now 600 ohm value) resistors wich is really hot. Is this the consequence of the higher T8 base voltage of 0.690v? Sould i drecrease it more?
What do you say should i go lower with these two resistors values or lower the 1k5 resistors more to get a lower bias?
I know that i have aprox 0.9mA across the "1k5" resistors wich is really really good but what about 0.379v across them? Is this voltage sufficient to keep the transistors open or should i readjust them to a higher ohmic value and readjust the other 180/240 resistors again to have 0.8-1 mA at aprox 0.6v?
Thanks
 

Attachments

  • IMG_3021.jpg
    IMG_3021.jpg
    740.6 KB · Views: 769
I want to werify the pcb, but i couldnt find any pcb error whatsoever (inversed transistors or wrong resistors). Can someone help me with a quick diagnose?
What do you say about the pcb? Does it has any error or ground loops?
This is the pcb that i use for the above schematic.
Please help!
 

Attachments

  • image.jpg
    image.jpg
    725.6 KB · Views: 709
Hi Sergio

You can not get the right voltage everywhere plus the right bias with out using the same rail voltage (Power supply voltage)
My advise go back to the original circuit as close as possible!!
Replacing components here and there will effect the performance of the amplifier to.
Only increase the 442R resistors to 560R or so
Please replace the Zener diode with lower value Zener.
For example
You have now 22V replace it with 20V and check the total bias! If it need you can go done to 18V to.
The small transistors get 1-2V less that cause less problem than you try to mess with all components (resistor values) here and there
For example here on these circuit they got 25V rail V, you can see the Diode value how low compare yours.
You asked me before how to reduce the Zener diode, you have to chose the diode V.
Greetings
 

Attachments

  • hiraga2fig1.gif
    hiraga2fig1.gif
    36.5 KB · Views: 686
Hi Sergio

You can not get the right voltage everywhere plus the right bias with out using the same rail voltage (Power supply voltage)
My advise go back to the original circuit as close as possible!!
Replacing components here and there will effect the performance of the amplifier to.
Only increase the 442R resistors to 560R or so
Please replace the Zener diode with lower value Zener.
For example
You have now 22V replace it with 20V and check the total bias! If it need you can go done to 18V to.
The small transistors get 1-2V less that cause less problem than you try to mess with all components (resistor values) here and there
For example here on these circuit they got 25V rail V, you can see the Diode value how low compare yours.
You asked me before how to reduce the Zener diode, you have to chose the diode V.
Greetings

Thank you Gaborbela. I'll try.
 
Hi guys. I want to build Jean Hiraga's Super Class A 20W Amp (L'Audiophile '34, Jan.1985), but I have some questions:
1) Is the sA 20W better than sA 30W?
2) Where is 2200 pF? It's not need in 20W version?

I have a small power supply 0,2F per channel, in future I'll make 0,75F per channel.
Transistors: 2SC1775E+2SA872E (in input), MJE15034G+MJE15035G or 2SC4793+2SA1837 (in drivers) and for output I have: MJL3281A+MJL1302AG, MJL21196G+MJL21195G, MJW21192+MJL21191G.
3) Maybe better to use 4 MHz transistors in output (if in schematic 2200 pF not exist)?
 

Attachments

  • Jean Hiraga Super 20W Class A Amplifier Schematic.jpg
    Jean Hiraga Super 20W Class A Amplifier Schematic.jpg
    206.3 KB · Views: 599
Hello my dear colleagues again.
I'm back and this time with good results:
The amp is working!!!
I have +-27v cc @1,65 A bias current.
It dissipates huge amount of heat after stabilisation after about 1 hour.
My problem WAS THE PCB. That's why i couldnt adjust the bias.
VERY IMPORTANT: dont design pcb by yourself if you dont know how the circuit works. It seems that this amp is the most sensitive amp that i have ever seen at this chapter............
Now i'm doing some listening tests, fine adjustings and listening to diffrent type of trannies for all stages from the amp. :D
Cheers
Sergiu
 
Hello my dear colleagues again.
I'm back and this time with good results:
The amp is working!!!
I have +-27v cc @1,65 A bias current.
It dissipates huge amount of heat after stabilisation after about 1 hour.
My problem WAS THE PCB. That's why i couldnt adjust the bias.
VERY IMPORTANT: dont design pcb by yourself if you dont know how the circuit works. It seems that this amp is the most sensitive amp that i have ever seen at this chapter............
Now i'm doing some listening tests, fine adjustings and listening to diffrent type of trannies for all stages from the amp. :D
Cheers
Sergiu

Good to hear it. What version you made? :)
 
Here are some thoughts from my recent experience with this amp:
1) the 20w sound more like a good tube amp but the 30w has some third harmonic distorsion in plus wich makes the amp desirable for other musical genders. If you like Diana Krall, Nina Simone, Chris Deburg and other similar singers then the 20watter and the "le monster" will blow you away..
The 30watter is more like First watt F5, you can listen to trance, country, vocal trance etc and (like the 20watter and "le monstre") you dont want to shut it down ever. :) I hope you understand what i'm saying here. In the end its all about tastes..
2) the 2200pf is paralelled with the 200ohm resistor from the output (in the 30w version) and that value is changed to 300 ohms.
I have seen allot of guys from a hungarian forum wich are using this amp in stereo mode in duall channel config with 0,1F per channel and said that this value is sufficient so no prob with 0,2F filtering.
2SC1775E+2SA872E (in input)-- is perfect BUT if they arent matched you are getting some distorsion at maxed volume AND it will be very hard to adjust the offset.
I'm using bc550/560c from NXP (ex Philips) wich are verry verry close to the ones that you want to use and are 1% matched.
At the input i recomend 1% matching or atleast match the trannies diagonally (ex 2x 2sa1% and 2x2sc 1% with a max of 5% between these pairs). I tried them both and worked like charm. ;)
For drivers i tested the following pairs:
a)2sa1837/c4793
b)2sa1220a/c2690a
c)2sa1930/sc5171
And found the followings:
a) the best pair, with a very precise sound, very good soundstage, pleasant mids, very good imaging, low bass and tight, extended hights and beautifull overall sound;
b) too fast pair because of the hight FT, flat soundstage (didnt have presence), tight bass but abit soft, the hights where not impressing and the mids where at the same level as the bass and hights so nothing impressing here.
The sound was focused but was hard to distinguish the instruments and effects and because these trannies where too fast it was hard and exausting to listen to the amp. It wasnt pleasant at all.
c) same as the b) variant;
Conclusion: altought all the above pairs where better aprox in all aspects compared to the original drivers used, Cob an FT, it seems that this amp doesnt love drivers with FT faster than 90-100 Mhz altought the CoB is better than the original ones....
I used as output pair 2sa1943/2sc5200 Toshiba NOS.
The source: ipod 5.5 classic@80gb@Wolfson DAC chip moded to Rockbox for FLAC format support.
Speakers: modified Tento closed enclosures rated at 40w sinus and 8 ohms from an Kashtan mag.
The music (in Flac format):
* luminary-amsterdam (smith and pledged remix); -- depht of field, vocals, effects;
*julio iglesias-crazy;-- vocal and for the big trompet (dont know the name in english);
* katie melua- nine million bicicle;--guitar, flaut, violin, vocals;
* sting-fragile;--- mandoline, dinamics, vocals, depht of field;
*dave brubeck-take five; dinamics, rithm, timbre;
*mantovani and his orchestra- limelight and moulin rouge;-- soundstage and precision;
*above and beyond--full concert film live from Porchester Hall (not FLAC; it is dirrectly from youtube@1080p); sound stage, sound stage and vocals;
I made the listening tests for 3 days from 8AM to 24:00PM continous listening to the same mellodies over and over again.... I think that after i'll finish the tests i will not want hear again those mellodies.. :D
So my friend i think that MJ15034 and his conplement and the 2sa1837/c4793 and other pairs with COB lower than 40pF and FT lower than 100mhz is best for this amp.
I would NEVER sugest 4mhz output pair for this amp because they are TOO slow; i would sugest atleast 10-12Mhz (minimum).
I have all the pairs that you listed in your post my friend and some sanken outputs and will listen to them too and will post my impressions again. I hope that at the end of the next week i will finish the listening tests... So dont rush to buy them as i did because you will loose some money on the road, so just be patience my friend.

I think i will start a new thread with secrets, typs and tricks for this wonderfull amp for others to find answers fast and easy.



My PSU is a dual mono 2x300w@2x27v cc@2x5A@2x 0,484F CrC psu per amp.
Now i am using an original pcb abit modified for my components and tastes.


Cheers
Sergiu
 
Hello! How you resolved an issue with the raised current of output transistors?

Hi Mihai,
I'll describe it in steps (for the 30watter versions):
a) first measure the 1k5 and 33k resistors to be matched (THIS IS VERY IMPORTANT);
b) measure the 500r trimmer; the value should be fix 500r if not put resistors in series one for each end of the trimmer (not the wiper) to give this value; --this will help you ballance ellegantly the offset if the input is matched;
c) USE THE ORIGINAL PCB--this is the most important thing---i used a self made pcb and couldn't adjust the bias;
d)replace the 180/240 ohm resistors with 500R trimmers and set them at the values from the schematic respectivelly 180/240 ohm;
e) short the input and no output load and set the 500r trimmer at half;
f) set offset: it has to be between +0,045v ~ -0,005v;
g) measure the voltage across the 1k5 resistors, if it isn't equal on both sides adjust from the "180/240" resistors values to have the same voltage across the 1k5 on both sides.. The voltage should be between 1,06v and 1,5v.
In my 30watter i have 1,078v across both of them and it works like charm.
h) readjust the offset in the same limit after adjusting the voltage across the 1k5 resistors;
i) now measure the voltage across the 0,33ohm resistors; you should have between 0,495v or UP if you have decent cooling;
If the bias is not it the range specified above, then put a 100k double pot in place of the 33k resistors set to 33k for start up and after start up adjust accordingly.
If you have higher voltage psu than +-25v (specified in the schematic) or above 27v cc, increase the 1k8/2w bias resistors. I have a 27V cc psu and it was not needed to increase the 1k8 value... If you go higher you have to increase this value accordingly.
DO NOT decrease the 1k8 value to adjust the bias because even if it will decrease the bias abit you will just increase the local NFB of the "darlingnot" output arrangement wich is not good. ---i tried it, it worked,but didnt listened to this mod...i reverted back to the original value;

***Another solution for bias adjust (if the above advices didnt work) is to lower in STEPS the 1k5 resistor value to minimum 680ohm (dont go lower), measure the voltage across it so that it will be in the parameters specified above (between 0,5v and 0,68v across it) and then readjust the 33k resistors to lower the bias..-- tried this version with succes.
Increasing the 0,33r to 0,47r value of the power resistors will also decrease the bias...
Hope it helps for new builders and i hope i didnt omitted any step.
Cheers
Sergiu
 
Last edited:
Hello! How you resolved an issue with the raised current of output transistors?

Hi, on my 30W version 0,63V (bias=1,57A across 0,4 om collector's resistors).
BJT:
2SC1775E+2SA872E, MJE15034G+MJE15035G, MJL3281A+MJL3281AG. (with Hitachi 1775+872 to my ears the sound a little better, than with ON Semi BC550CG+BC560CG).

p/s: using Philips BC550C+ON Semi BC560CG was bad idea.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.