FBJ
If you are considering the 'JLH for ESL' version for use with normal speakers, the supply rail voltage and quiescent current will need to be optimised to suit the speaker's impedance.
A number of people have contacted me about doing this, but so far the only completed project that I know of is Nick Gibb's original.
As a rough guide, given adequate heatsinking, it should be possible to achieve around 40Wrms into either 4 or 8 ohm or 30Wrms into both 4 and 8ohm by the appropriate choice of rail voltage and Iq.
Geoff
If you are considering the 'JLH for ESL' version for use with normal speakers, the supply rail voltage and quiescent current will need to be optimised to suit the speaker's impedance.
A number of people have contacted me about doing this, but so far the only completed project that I know of is Nick Gibb's original.
As a rough guide, given adequate heatsinking, it should be possible to achieve around 40Wrms into either 4 or 8 ohm or 30Wrms into both 4 and 8ohm by the appropriate choice of rail voltage and Iq.
Geoff
JLH
Geoff, I will refer to the schematic I have included:
There seems to be a bit of ambiguity as to the function of Tr2 and Tr5. This is caused be JLH's statement: "the design gave a somewhat lower distortion if the hFE of Tr1 was greater than that of Tr2. This caused the output circuit to act as an amplifier with an active collector load rather than an output emitter follower with an active emitter load." Are these transistors intended to add to the current gain of the output stage or are they supposed to act as a current source. I see in your version for the Quad ESL the two 2N3055s are most definitely configured as current sources.
Thanks,
H.H.
Geoff, I will refer to the schematic I have included:
There seems to be a bit of ambiguity as to the function of Tr2 and Tr5. This is caused be JLH's statement: "the design gave a somewhat lower distortion if the hFE of Tr1 was greater than that of Tr2. This caused the output circuit to act as an amplifier with an active collector load rather than an output emitter follower with an active emitter load." Are these transistors intended to add to the current gain of the output stage or are they supposed to act as a current source. I see in your version for the Quad ESL the two 2N3055s are most definitely configured as current sources.
Thanks,
H.H.
Attachments
Good eye Harry, I didn't notice the CCS at first look (I'm at work).
Thanks again Geoff- After building low power (Pass) amplifiers I think I will stick with the lower power JLH. With the right speakers these small amps can really sings and they are fun too. I got enough BIG *** AMPS already.
Thanks again Geoff- After building low power (Pass) amplifiers I think I will stick with the lower power JLH. With the right speakers these small amps can really sings and they are fun too. I got enough BIG *** AMPS already.
I don't know, I turn my back for half an hour and there are yet more questions to answer But in which order should I take them?
Yes, Jam, you are right about C4. Harry has used the diagram from the original 1996 Electronics World article which was incorrect. I added a text note under the diagram when I posted the copy on my website pointing out the correct connection.
Yes, Nelson, the DoZ does bear a strong resemblence to the JLH. When Rod was developing his BJT alternative to the Zen he found that the best solution was to revert to a 30 year old design.
Harry, as I mentioned earlier in this thread, I believe that the comment you quote from the 1996 article must be read in conjunction with Table 2 in the 1969 article. The measurements that JLH made on the original 1969 amp clearly show that the best distortion figures occured when the output transistors had a matched and relatively high gain. JLH realised that, at the time the article was published, many constructors would not have the luxury of matching output devices due to availability/expense. He therefore tested transistors with different gains and found that lower distortion was produced if Tr1 had a higher gain than Tr2.
I will now throw the cat amongst the pigeons and say that simuluation indicates that a lower THD is achieved by having a higher gain for Tr2, though matched devices still give the best figures. So far I have preferred to believe JLH's measurements rather than simulation but if anyone has the necessary test gear to check out which is correct I would be most interested to hear the results.
The output stage of all versions of the JLH is basically that shown below. I have always been in somewhat of a quandry as to how to interpret the method of operation. Tr3 clearly acts as both a voltage gain stage and a phase splitter. Doug Self describes the circuit as Tr1 being a voltage controlled current source with Tr2 being the current amplifier. However, JLH refers to an amplifier (Tr1) with an active collector load (Tr2). Perhaps someone who has had some formal electronic education can enlighten me?
You mention an ambiguity regarding Tr5. All this is doing is replacing the constant current source with a current source that is set by the output stage quiescent current. This removes the output stage quiescent current dependance on output transistor gain which of course will vary with current and temperature.
Geoff
But in which order should I take them?Yes, Jam, you are right about C4. Harry has used the diagram from the original 1996 Electronics World article which was incorrect. I added a text note under the diagram when I posted the copy on my website pointing out the correct connection.
Yes, Nelson, the DoZ does bear a strong resemblence to the JLH. When Rod was developing his BJT alternative to the Zen he found that the best solution was to revert to a 30 year old design.
Harry, as I mentioned earlier in this thread, I believe that the comment you quote from the 1996 article must be read in conjunction with Table 2 in the 1969 article. The measurements that JLH made on the original 1969 amp clearly show that the best distortion figures occured when the output transistors had a matched and relatively high gain. JLH realised that, at the time the article was published, many constructors would not have the luxury of matching output devices due to availability/expense. He therefore tested transistors with different gains and found that lower distortion was produced if Tr1 had a higher gain than Tr2.
I will now throw the cat amongst the pigeons and say that simuluation indicates that a lower THD is achieved by having a higher gain for Tr2, though matched devices still give the best figures. So far I have preferred to believe JLH's measurements rather than simulation but if anyone has the necessary test gear to check out which is correct I would be most interested to hear the results.
The output stage of all versions of the JLH is basically that shown below. I have always been in somewhat of a quandry as to how to interpret the method of operation. Tr3 clearly acts as both a voltage gain stage and a phase splitter. Doug Self describes the circuit as Tr1 being a voltage controlled current source with Tr2 being the current amplifier. However, JLH refers to an amplifier (Tr1) with an active collector load (Tr2). Perhaps someone who has had some formal electronic education can enlighten me?
You mention an ambiguity regarding Tr5. All this is doing is replacing the constant current source with a current source that is set by the output stage quiescent current. This removes the output stage quiescent current dependance on output transistor gain which of course will vary with current and temperature.
Geoff
Attachments
formal electronic education
I thought I had one till I looked at this circiut...... I wonder if a shootout between a version of this circuit with a true CCS and a true phase spliter quasi-comp output stage is in order. As the circuit stands I find it problematic in terms of output device matching and repeatable performance. I actually wonder if Mr. Hood truly understood the design.
H.H.
I thought I had one till I looked at this circiut...... I wonder if a shootout between a version of this circuit with a true CCS and a true phase spliter quasi-comp output stage is in order. As the circuit stands I find it problematic in terms of output device matching and repeatable performance. I actually wonder if Mr. Hood truly understood the design.
H.H.
I thought I had one till I looked at this circiut......
That's a pity, I'd hoped you were going to enlighten me! I have another reference to this topology, gleaned from Wireless World in the early '70s, which stresses the need for matched output devices and which has one or two interesting variations. One of the variations is a very simple, three transistor, Class-A, unity gain op-amp power booster.
I'm sure the performance is repeatable since many thousands of these amps must have been built by now. Since setting up my website, I have yet to hear anything other than favourable comments about the sound quality, even when compared with the likes of the Hiraga, Leach, Zen, Aleph and many highly regarded commercial designs.
Geoff
That's a pity, I'd hoped you were going to enlighten me! I have another reference to this topology, gleaned from Wireless World in the early '70s, which stresses the need for matched output devices and which has one or two interesting variations. One of the variations is a very simple, three transistor, Class-A, unity gain op-amp power booster.
I'm sure the performance is repeatable since many thousands of these amps must have been built by now. Since setting up my website, I have yet to hear anything other than favourable comments about the sound quality, even when compared with the likes of the Hiraga, Leach, Zen, Aleph and many highly regarded commercial designs.
Geoff
H.H.
Do not be offended if I ask you if you fully understand mr. hood's 96 version.
B.t.w. did you have a chance to read mr. hood's more recent publications? (various Electronics World 2000, 2001 issues)
To me mr. hood seems quite knowledgeable.
Could be I am totally wrong.
Maybe you can suprise us with your own killer power amp design?
Regards
Do not be offended if I ask you if you fully understand mr. hood's 96 version.
B.t.w. did you have a chance to read mr. hood's more recent publications? (various Electronics World 2000, 2001 issues)
To me mr. hood seems quite knowledgeable.
Could be I am totally wrong.
Maybe you can suprise us with your own killer power amp design?
Regards
Just passing through...
The driver/gain stage in Geoff's post is what a tube critter like myself would call a split-load phase splitter. They give virtually perfect matching of levels between + and - outputs, but my objection has always been that the Zout for the two halves differs widely; the top coming from the plate/collector and the bottom coming from the cathode/emitter (hence a follower). The difference in Zout can lead to disparities in frequenciy response between the top and bottom halves of the circuit at higher frequencies.
In other words, there ain't nuthin' new there...tube guys been doing it for years, albeit without the current source up top (usually--seem to recall seeing one or two of those, too...).
Grey
The driver/gain stage in Geoff's post is what a tube critter like myself would call a split-load phase splitter. They give virtually perfect matching of levels between + and - outputs, but my objection has always been that the Zout for the two halves differs widely; the top coming from the plate/collector and the bottom coming from the cathode/emitter (hence a follower). The difference in Zout can lead to disparities in frequenciy response between the top and bottom halves of the circuit at higher frequencies.
In other words, there ain't nuthin' new there...tube guys been doing it for years, albeit without the current source up top (usually--seem to recall seeing one or two of those, too...).
Grey
Speaking of the 7815 issues...
Well, i'm almost done with my JLH (finishing a beautiful wood enclosure, you'll drool at the pictures! ), and i have to build the "final" version of the amp boards; this is done on veroboard, as it's cheap, easy and i had no problems whatsoever with my previous prototypes.
So, has anyone tried the ccs mod instead of the voltage regulator & resistor? I never had problems with the 7815 (i increased the input and ouput capacitance and added a 4,7k resistor to ground to avoid any oscilations, and it worked just fine), but i feel the ccs Geoff proposed will perform much better, and the idea of eliminating those ics from the board is appealing to me. I recall Mr. Hood proposed the 7815 to eliminate noise and hum getting to the input stage bjt. Is this a concern with the proposed ccs?
Well, i'm almost done with my JLH (finishing a beautiful wood enclosure, you'll drool at the pictures!
), and i have to build the "final" version of the amp boards; this is done on veroboard, as it's cheap, easy and i had no problems whatsoever with my previous prototypes.So, has anyone tried the ccs mod instead of the voltage regulator & resistor? I never had problems with the 7815 (i increased the input and ouput capacitance and added a 4,7k resistor to ground to avoid any oscilations, and it worked just fine), but i feel the ccs Geoff proposed will perform much better, and the idea of eliminating those ics from the board is appealing to me. I recall Mr. Hood proposed the 7815 to eliminate noise and hum getting to the input stage bjt. Is this a concern with the proposed ccs?
Grey, thanks for passing through. I was aware that the topology was based on valve practice and this is hardly surprising since JLH cut his teeth on valves. But is this circuit a common emitter stage with an active collector load, an emitter follow with a variable current source or what?
Lisandro
The ccs replacement for the 7815 has been tried in a 1996 version as well as the 'JLH for ESL'. The feedback I have received is that it resulted in greater clarity and smoothness which is what I expected as regulators do inject high frequency noise.
JLH himself uses a ccs in another design that has a single input transistor and dual supply rails which is why I originally suggested that this arrangement be tried.
The ccs has reasonable supply rail rejection and there has been no problems with hum/noise so far, though the amps using it have had regulated supplies. I would expect more breakthrough from the negative rail than through the ccs. If you are concerned, the supply rail rejection of the ccs can be improved by splitting R3 ('JLH for ESL' schematic) into two and decoupling the mid-point to the +ve rail with a 47uF capacitor.
Geoff
Lisandro
The ccs replacement for the 7815 has been tried in a 1996 version as well as the 'JLH for ESL'. The feedback I have received is that it resulted in greater clarity and smoothness which is what I expected as regulators do inject high frequency noise.
JLH himself uses a ccs in another design that has a single input transistor and dual supply rails which is why I originally suggested that this arrangement be tried.
The ccs has reasonable supply rail rejection and there has been no problems with hum/noise so far, though the amps using it have had regulated supplies. I would expect more breakthrough from the negative rail than through the ccs. If you are concerned, the supply rail rejection of the ccs can be improved by splitting R3 ('JLH for ESL' schematic) into two and decoupling the mid-point to the +ve rail with a 47uF capacitor.
Geoff
Geoff said:
I thought of something similar too... actually, my supply is a fairly nice working cap-x, so hum is not much of an issue, but it always surprised me a bit that Mr Hood decided to place a regulator there when there're better (IMHO) options, like the discrete ccs.
I'll try the mod and get back to you when this amp is (finally) complete. Thanks again for the feedback!