As far as construction goes the old units definitely are better. Point to point wiring beats PCB's any day of the week 😉. I have repaired a few of the old units by now. They are very easy to service and am impressed by the layout. Getting at any of the components is very simple, as well as following the layout against a schematic.
https://www.youtube.com/watch?v=2HgS6gvokEI
The new MC275 units are all PCB construction. Their modern transformers look to have plastic bobbins
. I do not know about the old ones though. I think the "alien glow" green LED look is appalling. Nothing beats the natural glow of tube heaters. Also, too many solid state components have been introduced into the new MC275. Servicing them looks to be a bear as well.
I do not know if this is true, but someone at the RIT technology showcase mentioned McIntosh now is sourcing their tubes from China, as well as other parts. Judging from their modern construction, they look to be "Chineseified" yet carry snob prices. Back in the day, if one saved for a bit, the average middle class Joe would be able to afford a decent system from them. Is this true, they are sourcing tubes from China?
I do agree with Alan0354 the sound is not the most impressive. The McIntosh engineering was designing for sterile, accurate performance (hence the bandwidth and very low distortion). This does not necessarely mean the amp will sound good, as the tone is influenced by far more many variables, plus gobs of subjectivity (some like the sound of carbons comps and old paper caps where others care less).
https://www.youtube.com/watch?v=2HgS6gvokEI
The new MC275 units are all PCB construction. Their modern transformers look to have plastic bobbins
. I do not know about the old ones though. I think the "alien glow" green LED look is appalling. Nothing beats the natural glow of tube heaters. Also, too many solid state components have been introduced into the new MC275. Servicing them looks to be a bear as well. I do not know if this is true, but someone at the RIT technology showcase mentioned McIntosh now is sourcing their tubes from China, as well as other parts. Judging from their modern construction, they look to be "Chineseified" yet carry snob prices. Back in the day, if one saved for a bit, the average middle class Joe would be able to afford a decent system from them. Is this true, they are sourcing tubes from China?
I do agree with Alan0354 the sound is not the most impressive. The McIntosh engineering was designing for sterile, accurate performance (hence the bandwidth and very low distortion). This does not necessarely mean the amp will sound good, as the tone is influenced by far more many variables, plus gobs of subjectivity (some like the sound of carbons comps and old paper caps where others care less).
Last edited:
https://www.youtube.com/watch?v=2HgS6gvokEI
The new MC275 units are all PCB construction. Their modern transformers look to have plastic bobbins
I watched the video, I personally have nothing against pcb as I designed a lot of pcb myself. I would use a thicker pcb so it is more rigid and more reliable. I would use 1/8" pcb. It does not cost more, just much more reliable.
I am surprised it uses E-I core. For this fancy level, I would expect to have C core like Lundahl.
Still, how can you have 7 tubes for a power amp?!! My experience with guitar amp is the more circuit the signal goes through, it just get worst, not better. Nothing beats the good old simple tweed amps from the older days that have no bells and whistles, signal goes through only two tubes before going to the LTP and power tubes. There is a certain sweetness that all the fancy modern amps with all the effect cannot touch. I imagine audiophile amp are like this. Hell, people even eliminates the tone stack all together!!!
They use 3 gain stages and then some to generate the necessary signal strength for the unity coupled circuit. It is similar to a cathode follower in that the driver circuitry needs to generate much of the up-front gain. The penalty, pre-driver distortion (it sounds very nasty in Hi-Fi) 😉. The McIntosh appears to use feedback to address much of this issue. I did not measure the amount of feedback used in my MC-30, perhaps someone can document this. The designers looked to have made the up-front driver as linear as possible without the feedback, then closed the loop. It would be an interesting exercise to see how linear the up-front driver circuity really is, swinging large signals. I myself do not have time to perform the exercise as I have so much on my plate.
Again, I agree with Alan0354 that no matter what, adding more stages reduces the net linearity. How can a gain stage distort a signal if it is not there? Also stability is compromised with more stages encompassed in a feedback loop. A seperate winding on the output transformer mitigates this issue, but compensation networks were added in many places to ensure stability. Various revisions of the MC-40 (and other amps as well) eliminated or changed the values of some of these components. This is damning evidence that Frank and his boys struggled with the "poles in the east 🙂".
Again, I agree with Alan0354 that no matter what, adding more stages reduces the net linearity. How can a gain stage distort a signal if it is not there? Also stability is compromised with more stages encompassed in a feedback loop. A seperate winding on the output transformer mitigates this issue, but compensation networks were added in many places to ensure stability. Various revisions of the MC-40 (and other amps as well) eliminated or changed the values of some of these components. This is damning evidence that Frank and his boys struggled with the "poles in the east 🙂".
Last edited:
As far as construction goes the old units definitely are better. Point to point wiring beats PCB's any day of the week 😉.
The new MC275 units are all PCB construction. Their modern transformers look to have plastic bobbins
Sometimes good point-to-point may be easier to service, but all advantages are end here. I have no problems to remove individual elements from the PCB.
And what's wrong with plastic bobbins? Generally, bobbin allows more accurate and reliable construction of wound elements. Using bare tube is not an advantage, its cheaper shortcut.
this is what the original opt back in the day....simply not available anymore...
Why would that be unavailable today to a company like McIntosh? You can get phenomenally good C cores, amorphous even, Metglas for example. I'm sure they'd make whatever McIntosh would need considering the number of units they could sell. Going to a cheaper core sounds more like cost-cutting to me.
There is nothing wrong with Chinese vacuum tubes. Especially if used in an amplifier that has loads of negative feedback like the McIntoshes all do. The sound is controlled by the compensation in the NFB loop more than by the components.
This is neither good nor bad. It is what gives this company it's "sound" You can like it or not. I prefer a more traditional and much simpler audio amp.
About the comment the PCBs were OK. Yes they work fine until they fail. A few years ago someone GAVE me a PCB based tube amp where some part failed and caused several PCB traces to melt and a passive part to actually catch fire. It was a chain reaction where one burning part melts the parts and traces near it. The charred mass was basically unrepairable. This would never happen with a hand wired amplifier. PCBs that are badly damaged are very hard to repair.
(My best guess is that this 100W amp was running at full volume and a speaker "blew" and cause an open circuit on the OPT secondary and the transformer then shorted. Some resisters arrow the the power tubes overheated, result was melted traced, burned fiberglass and resisters acting like fuses.)
My preferred construction method is turret boards. They are bomb proof and good enough even for music instrument amps they get loaded into vans and hauled up and down stairs on dollies hundreds of times for decades.
I agree there is nothing wrong with Chinese tubes. I use them and find many of them to be quite good. What I am getting at (I should have been more clear in the first place) is that McIntosh is traditionally an American company, and selling out to China may harm their image.
Tubes are not made in the US anymore so one can argue they have little choice.
Tubes are not made in the US anymore so one can argue they have little choice.
That's where good pcb layout technique comes into play. More often, people use too small a trace, when they use via to go from top to bottom, they use too small a via that it becomes the hot spot. It's all about the layout.You can make pcb as robust or even more than hand wire. The strong point about pcb is the layout is much more consistent, you don't worry about one wire in the wrong place. I can even put ground plane. With good layout (if), pcb out perform hand wire any time of the day. The problem with McIntosh to me is they worry too much in how it looks, they put the tubes up front to show, they are hand tied to make it look good. Just looking at the video from yesterday, the first stage and differential stage has to go on a straight line parallel to the front panel, then drive the power tube at the back. The signal path has to turn a right angle or worst, even have to turn back. The best is to go in a straight line. Maybe, that's the selling point that people like the looks.
Labor is very expensive to hand wire now a days, pcb does save a lot of cost in assembling.
Making the pcb of 1/8" can be very strong, not easy to bend and crack.
There is psychological element in this that pcb is cheap.
Last edited:
As part of my normal work, I design PCBs all the time. In high heat applications, I select the proper material. For high voltages, I follow IEC guidelines. I even exceed them if possible. If I have to place power output tubes on a PCB, I always use the Chassis as a heat shield.
As a rule, I try and avoid high voltages on multilayer boards and running them through vias. For my tube analyzer, I set aside real estate for high voltage and used optocouplers to bridge over to the digital logic 4-layer portion.
I study the specs very carefully, plan my layouts, anticipate operating temperatures, and even figure insulation breakdown in my designs. They don't teach you in school how to deal with 600 volts in a PCB. I had to learn it on the job.
As a rule, I try and avoid high voltages on multilayer boards and running them through vias. For my tube analyzer, I set aside real estate for high voltage and used optocouplers to bridge over to the digital logic 4-layer portion.
I study the specs very carefully, plan my layouts, anticipate operating temperatures, and even figure insulation breakdown in my designs. They don't teach you in school how to deal with 600 volts in a PCB. I had to learn it on the job.
I was really impressed by the awesome "simple line stage" you posted last week.
http://www.diyaudio.com/forums/tubes-valves/265913-simple-line-stage.html
Art
http://www.diyaudio.com/forums/tubes-valves/265913-simple-line-stage.html
Art
Here is video from a factory here in California where they are building hand wired tube amplifiers. Yes hand wired in the US.
The circuit boards are NOT PCBs. There are no traces on the board. It is 1/8th fiber board that is drilled then they install eyelets, just like the kind used for shoes like Converse tennis shoes. You use either a mallet or press to set these. The eyelets are made of tin plated brass. The component and wire are placed in the eyelets and soldered. The board is just a mechanical device.
http://youtu.be/54sNv1Ck6dc
The chassis layout minimizes the length of the wire runs. Passive parts are inside, tubes on the bottom, jacks and controls on top and transformers on the rear. All three sides and the inside of the U-shape classes are used for mounting stuff. It is compact but also very easy to work on. No need to turn anything upside down for access.
I built a clone of this a while back. It is very easy to service and can take loads of physical abuse. One modification I did was to use a lot of colored coaxial wire. Using shielded wire kept noise floor down.
ChampRear.jpg
I actually am building a fully balanced transformer coupled version of that simple line stage right now as we speak.
Last edited:
If I were going to do something "along the lines" of the Macs, I would first try to separate out the useful innovations from the pointless ones. The unity coupled circuit eliminates inductive discontinuity spiking for class B. But no one in their right mind would use class B for an audio amplifier. The unity coupled circuit does nothing for eliminating crossover distortion itself, except by their adding on gobs of NFDBK after the fact.
The high 50% CFB makes for linear output tube and OT performance, but at a steep penalty for drive circuitry. And the unity coupled (bifilar primary to primary) OT effectively reduces the # of turns by 1/2 to improve OT bandwidth. But it seems they just used that to enable cheap random wound OTs (at least later).
So lets start out with the "right stuff". Class AB, high current tubes for low primary Z, toroid or at least cut core OT. The important thing about CFB is the close coupling with the output, especially for the cathode coupling which looks like a voltage source.
So, lets say we use a 10:1 OT with 10% bifilar CFB (but bifilar with the secondary here). Another 10% of the primary around the B+ tap can also be bifilar with the secondary, so we end up 10% trifilar (actually 5-filar considering both sides of the P-P, so this might be wound as a twisted bundle). (and use good wire insulation, some super wire coatings available now) The 10% CFB and 10% around the B+ can be cross cap coupled as well. So effectively this is giving 20% bifilar between total primary and secondary.
Now the other 80% of the primary (each P-P side) is being driven by a pentode plate, which looks like a current source. A current source goes through any leakage inductance unhindered (as long as there is sufficient plate voltage compliance). So the winding technique for the remaining 80% plate winding sections can be designed with robust insulation, to eliminate distributed capacitance (bad stuff for current sources), and relaxed concern for low leakage L.
Since the 10% bifilar CFB section(s) are so tightly coupled to the secondary, we will also use it for feedback to the differential driver circuitry cathodes to make it more effective. We will also want to smooth out the output tube gm across the class AB range better for minimal crossover distortion. So we make sure the CFB sections have equal winding resistance (hidden current sense resistance) for differential current sensing by the differential driver besides the usual inherent voltage feedback. (possibly adding additional Rsense in the output cathode circuits) The differential driver also gets a selected tail resistance for removing the 3rd harmonic component.
(differential current between the output tubes is the same as actual output current, so by making it track the input signal as well, we make the gm stay constant across the crossover region) Differential current feedback not only irons out the gm variation, but in turn makes the output Z (damping factor) constant across the signal range too. (before adding global feedback)
Front end can be most anything suitable. Global feedback can be up to what the OT will conveniently allow (OT bandwidth should be pretty good with the prescribed approach). Fill in the remaining blanks......
The high 50% CFB makes for linear output tube and OT performance, but at a steep penalty for drive circuitry. And the unity coupled (bifilar primary to primary) OT effectively reduces the # of turns by 1/2 to improve OT bandwidth. But it seems they just used that to enable cheap random wound OTs (at least later).
So lets start out with the "right stuff". Class AB, high current tubes for low primary Z, toroid or at least cut core OT. The important thing about CFB is the close coupling with the output, especially for the cathode coupling which looks like a voltage source.
So, lets say we use a 10:1 OT with 10% bifilar CFB (but bifilar with the secondary here). Another 10% of the primary around the B+ tap can also be bifilar with the secondary, so we end up 10% trifilar (actually 5-filar considering both sides of the P-P, so this might be wound as a twisted bundle). (and use good wire insulation, some super wire coatings available now) The 10% CFB and 10% around the B+ can be cross cap coupled as well. So effectively this is giving 20% bifilar between total primary and secondary.
Now the other 80% of the primary (each P-P side) is being driven by a pentode plate, which looks like a current source. A current source goes through any leakage inductance unhindered (as long as there is sufficient plate voltage compliance). So the winding technique for the remaining 80% plate winding sections can be designed with robust insulation, to eliminate distributed capacitance (bad stuff for current sources), and relaxed concern for low leakage L.
Since the 10% bifilar CFB section(s) are so tightly coupled to the secondary, we will also use it for feedback to the differential driver circuitry cathodes to make it more effective. We will also want to smooth out the output tube gm across the class AB range better for minimal crossover distortion. So we make sure the CFB sections have equal winding resistance (hidden current sense resistance) for differential current sensing by the differential driver besides the usual inherent voltage feedback. (possibly adding additional Rsense in the output cathode circuits) The differential driver also gets a selected tail resistance for removing the 3rd harmonic component.
(differential current between the output tubes is the same as actual output current, so by making it track the input signal as well, we make the gm stay constant across the crossover region) Differential current feedback not only irons out the gm variation, but in turn makes the output Z (damping factor) constant across the signal range too. (before adding global feedback)
Front end can be most anything suitable. Global feedback can be up to what the OT will conveniently allow (OT bandwidth should be pretty good with the prescribed approach). Fill in the remaining blanks......
Last edited:
That figure is A-weighted wrt a level of 10mV. The A-weighing means that all the LF noise is filtered out before measurement.
The 10mV reference level is a very high level not representative of actual use conditions. No wonder the noise measures -80dB.
Last edited:
Really?
I suggest that anyone interested in the Mac power amp approach read "Realistic Audio Engineering Philosophy" by Crowhurst. Those amps had piles of clever design elements. In a similar vein, study the schematic of Tim deParavicini's Luxman MB3045.
These were some damn smart engineers.
BTW, the vast majority of that preamp circuit was controls and displays; it's really a control center, not just a simple preamp. The audio circuits are indeed quite simple and are basic variants of the old Dynaco/Marantz circuits of the 1950s.
i have not heard the 275 play, but i have seen them before and those were not working units...
for all the complexity of that amp, and according to what i read in many places, some liked the amp, while still others were unimpressed....
my experience so far....even my concertina driven 832 pp amp can sound fabulous all around by those who heard the amp...the simplest of tube circuits can really amaze you.... 😀
for all the complexity of that amp, and according to what i read in many places, some liked the amp, while still others were unimpressed....
my experience so far....even my concertina driven 832 pp amp can sound fabulous all around by those who heard the amp...the simplest of tube circuits can really amaze you.... 😀
- Status
- Not open for further replies.