alzinski

You need to know how to interpret the data presented. I will explain my understanding and reasoning for not changing out the old Aerovox caps. When time permits I can do an actual write up in a separate thread if anyone is interested. I didn't use an LCR meter I used a signal generator and an oscilloscope. Let us talk about the 2uF caps first. I forget where the actual crossover point for the tweeter is in the AA crossover but I will use arbitrary numbers for the example. The tweeter is only handling the higher frequencies, looking at the ESR of the 2uF cap at 100Hz tells us nothing. In fact higher ESR at this frequency will only help us IMHO because it will attenuate more lows going to the tweeter. The Xc (capacitive reactance) of a 2uF cap at 2kHz is 40 ohms. Do you really think a cap that has an ESR of .4 ohms at 2kHz is going to make a difference? It won't. ESR will decrease with increase in frequency so test with the lowest frequency for the application. When testing the 13uF cap you may want to now test down to 500Hz. The Xc of a 13uF cap at 500Hz is 24 ohms. Do you really think .5 ohm of ESR is going to make a difference? I don't. The actual value of capacitance will make more of a difference. For example if the cap actually measured 17uF now the Xc is 18 ohms which will change the crossover point much more than .5-1 ohm of ESR. I think the best way to test these caps is with a signal generator and an oscilloscope at the lowest frequency for their application. The output amplitude will be a result of the combination of BOTH ESR and capacitance without having to worry about either separately. Use a series resistance with your signal generator and measure the amplitude across the cap. To summarize what I am talking about and what I did; A perfect 2uF cap at 2kHz should be 40 ohms. Take a 1v signal and shoot it through a 1k resistor in series with the cap. You should measure ~40mV across the cap. If you don't have ~40mV either the capacitance is off or the ESR is off or both. A perfect 13uF cap should be 24 ohms at 500Hz. Shoot a 500Hz signal though the same 1k series resistor and measure across the cap. It should read ~24mV. If you don't get 24mV across the cap either capacitance is off or ESR is off or both. I hope this makes sense and helps.

Great Post!!! We all know most recording "engineers" are just wannabee musicians that can't play an instrument.

Yes a perfectly balanced push pull amp is symmetrical which produces odd harmonics (for that stage only, it won't cancel anything before it). Lke I said before it will never be perfectly balanced (exact mirror images of each other) so there will still be some even harmonics but they will be much lower than a typical single endeds stage.

I found that with many reviewers they have no idea what they are talking about technically. I just want to clear up some bad information regarding push pull amps. The author makes it sound like you physically cut the signal in half into "plus and minus'" and then "recombine" them. Eh, that's not how it works, it's an AC signal, there is no plus or minus as it alternates hence the name. A push pull amp instead will make a mirror image of the signal that is 180 degrees out of phase, so at any point in time one signal is in it's positive phase while the other is in it's negative phase, the potential between the two phases is greater just like your house power, you measure from any phase to neutral you get 120v but measure phase to phase you get 240v. If you are trying to get more power to the speaker instead of having one tube swing 800v on the primary you have two tubes that swing 400v out of phase which will give you 800v. Having a push pull amp does not automatically mean you have crossover distortion, it can be run Class A which means no active device goes into cut-off; i.e. all devcies conduct current the full 360 degrees of the input waveform. A single ended amp has to conduct the full waveform so naturally it's going to be Class A. Crossover distortion is when an amp runs into Class B operation and one of the devices is in cut-off (non conducting) for longer than half the cycle (greater than 180 degrees), this will give a little notch at the zero crossing henc the name crossover distortion. The issue with push pull amps is the non-linearities naturally inherent with the devices and circuits won't exactly be the same for both phases, so they will never be an exact "mirror image".

You currently have an integrated amp. I assume you do not already have a preamplifier? The Horn monos have very little gain and require a 6v input to reach rated output power. i.e. You are going to need a preamp with voltage gain to drive them.

Ahhh I gotcha. Thank you for clarifying.

Testing of said amplifiers is done with resistive loads. Not all amps will behave the same way when a capacitive load is introduced like in the real world, some behave badly giving way to instability and ringing. This will be audible. SS gear requires a substantial amount of negative feedback to linearize the active devices and get distortion acceptable. Introduce large transients or worse clipping into high feedback amps and they all don't behave the same. There will be time domain distortions E.G. Transient Intermodulation Distortion. Devices and or circuits with low slew rate and high amounts of feedback will be the worst offenders. So even though "specs" on paper look good and you would think they should all sound the same, in the real world with real loads and real music going through them they do not behave the same and hence do not sound the same. I will agree that speakers and room play a very large role in the total equation, but it's foolish to overlook amps because certain specs appear to be similar on paper.

The large 12uF I measured .4 ohms. The smaller 2uF caps measured .5 and .6 ohms. For kicks I measured some bipolar electrolytics that are much smaller and brand new, they were ~2 ohms. The capacitance has drifted off though, I measured 17uF for the large one (should be 12uF) and 3uF for the smaller ones (should be 2uF). Not sure how to proceed. I was more worried about the ESR than the capacitance but now I am thinking the crossover points have been lowered. I remember reading somewhere that people actually mod the squawker to cross a bit lower, like ~400Hz?

If tapping the chassis makes the noise then tapping the tubes will most likely also make the noise (even if it's not them) and might not be the best method of diagnosis. I suggest swapping the two phono 12AX7's with two other 12AX7's and see if the noise changes or goes away. Ceramic capacitors have piezo electric properties, if changing the tubes around doesn't help you or tell you anything you may need to poke some of the ceramic caps with a chopstick.

I am first and foremost a musician, I have been playing and studying music for 40+ years. I have gone through extensive music theory and ear training. I am extremely emotional and passionate about music, I am music and I would die without it. Let's not confuse technical matters with emotional matters. I like what I like, you like what you like, I tend to think we probably like the same things. BUT people take it very personal when discussing technical matters, I don't know why.

Typical school teacher these days, everyone is a delicate flower, a unique snowflake, everyone gets a sticker and a pass grade for trying. Do you need a safe space because you didn't hear what you wanted to hear? I never said active line stages are NOTHING BUT distortion and noise, my point was that they can only add more noise and distortion than a simple resistive divider, this is fact and sorry if it offended you. I am not against active line stages at all, I never once said they sounded "bad". In some cases where you have a very low ouput impedance source and very high input impedance amp, and you run very short interconnects a passive volume control can work just fine and be very transparent. In some cases it won't work and you will need an active buffer or even something with some gain. The definition of high fidelity gets lost in these discussions, the device with the least coloration and which is the most authentic to the original material is the most hifi, whether you like it or not doesn't matter. If you have been reading, I myself use a SET amplifier and I don't for one second say it's high fidelity, I don't claim it has zero distortion, but I do like the sound. I never once said the amps being discussed in this thread sounded bad either. Just like the other Scott 299D thread the complaints were aimed more at the marketing aspect which in the audio business is most of the time a joke. My only complaint about the amps in question here is the marketing. My point was that they are intended to be used with an active preamp that has gain. The noise measurment they give for their Horn amplifier isn't impressive given the gain of the device and once you plug a preamplifier into it the noise floor will increase. It's my opnion that it's misleading, that's i!. If you have the horn amplifier by quicksilver hooked up to an efficient horn speaker and preamp, I suggest turning the volume up without any music being played through it and put your ear up to the speaker, I bet you will hear a little noise increase as you increase the volume. Again this doesn't make the amplifier a "bad" amplifier, it doesn't mean it will sound "bad", it doesn't really mean anything except the fact the marketing in my opinion is a joke. Some people don't understand specs and how gizmos work, they may even believe that using this amp no matter what preamp they use will have a 120db noise floor, but it's dependant on preamp being used, especially since they run unbalanced inputs. A critique on the amp itself would be something like this. Since the amp is intended to be used with an active preamp, it would be much more advantagous to use an input transformer to galvanically isolate it and break ground loops, this also has the benefit of being able to accept balanced inputs which will be quieter due to common mode rejection, and it can accept unbalanced sources, this also gives you the benefit of using all differential circuitry. But good transformers are expensive and they can impart their own distortion and limit bandwidth so everything is a tradeoff. If you post about your equipment on the internet you will have to accept all sorts of critics not just the good ones. I am not bashing the amp, I bet it sounds fantastic and I hope the owner is very happy with it. But this is a discussion about the product and as long as things are kept factual I don't see a problem with the truth.

I had to look at the manufacturer web page to see what we were discussing exactly, I had a good laugh. So 6 volts in gets you 25 watts out, that's a gain of 2.4. A typical source of 1v doesn't even get you to 1 watt of power with that amp. Why make a 25 watt amp to only get 1 watt of power? Seems like a waste to me. IF you want to be able to reach the amps full potential power then you must get some gain from somewhere because most sources do not put out 6v. It's really not that impressive at all considering what it is. To be more transparent with their specs, they should do a noise floor measurement with an unbalanced preamplifier hooked up to the input, that way there you will see the real world noise including ground loops, I am sure it will be substantially higher.

If you ask me having a power amp that asks 6v to reach full output power is wimping out and passing the buck in order to boast low noise specs. I do get it, when you need the VAS stage of a power amp to to swing 140v peak to peak to drive the finals it can send some "engineers" running. FIrst order of defense is good filtration in the power supply and second is front end stages that have good power supply ripple rejection. I personally like active loads like a constant current source, you can reach near full mu of a valve along with getting excellent linearity and power supply ripple rejection. A good one can present a near infinite load (which would look horizontal for those who are into drawing load lines) and not drop much DC across unlike a high value resistive load. I have heard many good DHT amps that were dead quiet on horn speakes that reached full output power with a normal input source; e.g. 1vrms. What Manley did was not considered good engineering in my book.

This pill I can swallow much more than the crossover distortion argument, I originally said I can sort of see doing it with certain dielectrics like aluminum electrolytics (these do have high DA). Teflon and PP caps have an extremely low DA and I would think being in a low Q circuit would dampen any artifacts, basically there is a low impedance path to earth through the voice coils of the squawker and tweeter. 12uF isn't a large value and most of us can spring for a decent cap at this value, when large value caps are needed >40uF I can see the use of certain dielectrics with poor DA and then battery bias could be worth it. For now I feel comfortable using a good film cap (possibly a motor run) for doing my cap refresh in my AA networks without any batteries. I am also toying with surplus military caps from Russia. Does anyone know the ESR range of the original caps that were used when they were new?

It wouldn't surprise that the ESR has increased which is why I want to change the caps out. As for the "charge coupled network", IF people are actually hearing a difference could it not be attributed to the two caps being put in series effectively increasing the ESR? From what I have read the battery bias is supposedly for getting rid of "crossover distortion". Stuart Yaniger couldn't measure any crossover distortion and so far in all my searching is the only person that has actually done any sort of testing. Here is the link: http://www.diyaudio.com/forums/multi-way/124824-electrolytics-sound-fine-post1543205.html#post1543205 If the battery bias trick actually does anything it should be measurable, and if anyone has any links they can share to actual measurements then I am going to assume it's a placebo effect, same as people saying they can hear differences with exotic speaker cables, interconnects, and power cables. Especially if you guys are doing this to film caps its a bit ridiculous.

I have asked folks about this in the past and not one can give me a half decent explanation. I can maybe see doing this with bipolar electrolytic caps (why anyone would use them to begin with is beyond me) but why would you do this with a film cap? I will try and dig an old email out for when I asked a chemist that I believe that worked in the capacitor industry, from memory I believe he laughed at the notion. I still plan to set something up to measure the ESR of the original caps. I will post the results for anyone interested.

Looks like a Hammond Enclosure.

Engineers have decisions to make and there are always tradeoffs. Benefits of Valve rectification is no switch noise which is found in SS diodes and most importantly is with inderectly heated valves you will get a controlled soft start. The latter is very desirable and removes the need for soft start circuits. The switch noise found with SS diodes can be mitigated via snubbers and or the use of Schottky diodes. The negative aspect of using valve rectifiers is the limited repetitive peak forward current rating, this limits the amount of capacitance and load that can be used. Using a choke is not an en vogue, it is necessary as Maynard pointed out. With a full wave rectifier you are trying to filter out 120Hz, you can view the filter as a potential divider. The choke or the resistor is the serial element and the capacitor is the shunt element in the filter, like a potential divider you want the serial element large and shunt element small; Vout = Vin * ( R2 / R1 + R2 ) The reactance of say 40uF at 120Hz is 33 ohms. To simplifiy I am going to ignore phase but this will illustrate my point just fine. To get any sort of filtering we would want to make the serial element (R1) much larger than the shunt (R2), at least 10x but the larger the better. Say we use a 330 ohm resistor; 33 / 363 = .09 We are left with 9% of the ripple so we either need to make R1 larger or R2 smaller, or both. But if we have say 20mA of load current and we want to increase R1 to 1k we will end up losing 20 volts of DC, it gets worse the more we increase R1 value. This wastes energy into heat across the resistor which is bad. The best way of doing this is to use a Choke which has a high impedance to 120Hz but passes DC, the only loss is the DC resistance in the winding. So let's use a 10H choke with a DC resistance of 150 ohms. The choke will have an impedance of 7.5k at 120Hz but we will only lose .02 * 150 = 3 volts across it. Lets see how much we reduce the ripple; 33 / ( 7500 + 33 ) = .0044 We are now left with .44% of the ripple and we only lost 3 volts DC across the choke. With SS rectifiers we can have much larger capacitance. Let's use 1000uF, this is 1.3 ohms at 120Hz. With a 330 ohm resistor we get .39% of the ripple and only lose 6vdc. So you can see with SS diodes you can use large values of capacitance and smallish value resistor so long as the load isn't huge (like with a preamp) and get very good filtration without spending money on chokes. SS diodes also have a very small forward voltage drop no mattter the load, usually .6v. A tube recitifier can have a large value internal impedance, this comes into play with amps that has a varying load like with Class AB amps, it adds compression. One way around this is to use a L input filter instead of a C input filter. I will save that rant for another time.

What's the power output? Pentode you will get a few watts, Triode will be maybe 1 watt. Japanese? I hope it has a bucking transformer in it because it was probably intended to run off 100VAC.

It is misleading, It would be great if Mat could ask the engineers some basic questions to why? I figured it's just a tip of the hat and to actually clone the 299d wouldn't be the best idea; I am no engineer but here is me best guess. First there are much more EL34 manufactureres out there to choose from and they are less expensive than the 7591's. The EL34 is a 25 watt tube where the 7591 is only a 19 watt tube. The two have similar transconductance but the 7591 has double the internal impedance. Running the pentode as a pentode will also have a lot more distortion along with the higher output impedance, this will require more negative feedback to lower distortion and output impedance. The original Scott used a small signal pentode for a gain stage before the triode split load phase inverter to get a higher open loop gain without adding another stage, more stages increases phase shift which leads to stability issues with large amounts of feedback. The 6U8 triode has high gain (40) further increasing open loop gain. Now running the output stage with local feedack as with distributed loading (ultra linear) you increae linearity lowering distortion and impedance, this in turn will require much less open loop gain for you won't be needing as much global feedback to clean things up. This is most likely why they didn't need a small signal pentode for high gain and a high gain triode, the 12AU7 will do. Personally regardless of what output device was used as I am okay with going with the EL34 because they are much more available, I would have stuck with Pentode mode as to retain the same sonic characteristic. Maybe they didn't want to fuss over getting the feedback loop stable? It can be a daunting task to get the feedback lead/lag compensation networks correct when using gobs of global feedback. At 1% THD it would appear they aren't using a lot of feedback if any at all so maybe they are only using one 12AU7 per channel for the power amp? I mean from reference my PP Class AB EL34 amp (ultra linear) from the fifties has .05% THD at full output and ~1% THD when run open loop. Maybe they aren't running any globabl feedback. Even still with one 12AU7 per channel it would be tough to drive the EL34's. One triode would have to be a split inverter, that would mean you would need to squeeze every ounce of gain from the other triode for full output, a CCS load would do that. This leaves one 12AU7 per channel for the phono stage, even if they did this would one 12AU7 be sufficient for a MM stage? It would take some tightrope walking but maybe. It's fun to speculate but even better if I could get my hands in one of them.

The amp probably doesn't have a snubber or flyback diode on the mains side of the power supply. When you switch the amp off the inductive load thows a huge voltage spike which you need to suppress. https://www.idec.com/language/english/AppNotes/Relays/contact_circuit_protection.pdf

Since this is a preamplifier we are discussing there won't be any sag due to the relatively constant load current. Usually output stages that go into class B and have unregulated supplies will get some sag and natural compression. Changing from one type to another, like 5Y3 to a 5AR4 can change the operating points by increasing or lowering the rail voltage.

There must be a communication breakdown for you are not understanding me. I am not arguing preference at all, we like what we like. What I am saying is it is a universal truth and a fact that an active gain device is going to add noise and distortion, significantly more than a resistor. Answer this question for me. Which is going to be more transparent? (i.e; Add less distortion, noise, phase shift etc..) A metal film resistor or a DHT line stage amplifier? You used the word "suffice". Ask yourself what goals you are trying to accomplish. For example, "how do I attenuate my CD players output to feed my power amplifier?". Does it make any sense to attenuate it down to millivolt levels, then use an active gain stage to bring it back up to an amplitude it was already at? Not really, and now it has more noise and distortion than it did to begin with. It seems extremely counterproductive to me and if it doesn't to you then you are kidding yourself. You may prefer one thing over another but it doesn't hide the facts of the matter, it adds noise and distortion.

Thank you for the background and the insight into the product. IMHO being assembled in America is much better than assembly in China. If anyone can point me to an amplifier where every piece inside it is made in America I will buy them a cigar. We live in a global market and if Germany is making a better capacitor for a specific application I will purchase it, to not because it's made overseas is doing yourself a disservice. This reminds me of the Carver challenge. Although the two amps may not have the same innards they may indeed have a similar sonic characteristic. You don't need to use the same topology to get the same in and out characteristics.

The voltage coefficient of resistors, especially metal film is so low you can't even compare the added distortion to that of what a directly heated triode will add from non-linearity. For people looking for accurate reproduction a resistive divider will be much more transparent. In your case where you need a buffer for cables the passive route isn't an option, but for some people it is an option and is also the most transparent.