Impedance matching / bridging between tube gear and modern audio interfaces

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Protecting humans from themselves is a full time gig for some safety agencies.

In audio there are considerations about speaker voltage that gets large enough to shock the meat puppets, but music not being continuous (mostly) is not much risk of humans getting stuck to a speaker lines.

JR
 
Protecting humans from themselves is a full time gig for some safety agencies.

In audio there are considerations about speaker voltage that gets large enough to shock the meat puppets, but music not being continuous (mostly) is not much risk of humans getting stuck to a speaker lines.

JR
They should put spikes in front of speaker stacks at concerts - stop the headbangers putting their heads into the speakers - like magpie spikes on a bike helmet.
 
Only power amps and long line drivers need to be impedance-matched to the load, because you want to make sure that maximum power is transferred to the load, which means maximum voltage AND maximum current.

With input audio devices, however, including outboard line-level gear, you are transferring voltage, not power (look up Ohm's law, voltage dividers). Here, the rule is that the input impedance of the next unit in chain needs to be considerably higher than the generating unit. For example our preamps (Sonic Farm Audio) have a mic input impedance switchable between 1, 2.4 and 10kΩ. There is no perceived level change when switching between them, only with the dynamic mics, whose impedance increases with frequency, there will be a little bit of a high end rolloff at 1k and 2.4k settings, which is the purpose of that switch.

Another example is an audio interface, any brand. As a rule of thumb, the line inputs on those are always solid state, no transformers, and most of them have an input impedance between 5 and 20kΩ. Engineers would go nuts if they had to impedance-match all those line-level units.

Tube stages do have a high output impedance, but the final output of those unit always has a low one, typically below 600Ω. This is achieved either by a step-down transformer or a solid-state buffer. I prefer the second solution, it's more elegant, preserves the tube sound and requires less gain at the tube stage.

One of the most important consideration of mic preamps is noise performance, which has its own impedance matching rules, but this is an issue that designers need to take care of, not the user. Typically, tube and FET stages have a much higher input impedance than transistors, and so mic input transformers in those preamps have a much higher transfer ratio (1:10 or even 1:20). Mic input transformers provide a totally noiseless gain, but they multiply the source (in this case mic) impedance by the square of the transfer ratio, so 100 to 400 times. The "source" impedance seen by the tube/FET will thus be (for a 200Ω mic) 20k or even 80k. Because you can easily achieve a tube/FET stage input impedance of 2.2MΩ or even higher, the signal transfer to the tube will still be optimum. Just don't overdo the transfer ratio, because the input capacitance of the tube/FET stage will start to roll off the high end. There you go, now you can even get into the preamp design.
 
Only power amps and long line drivers need to be impedance-matched to the load, because you want to make sure that maximum power is transferred to the load, which means maximum voltage AND maximum current.
Modern solid state power amps are voltage sources that exhibit low output impedance relative to speaker loads, so are effectively bridging outputs (load is nominally 10x source impedance). Vacuum tube power amps exhibit higher output impedance than solid state so use transformers to better match speaker loads to the power tube's source impedance.

Long lines are best driven from source impedance equal to their characteristic impedance to reduce reflections caused by mismatched impedance terminations at junctions. Such reflections can corrupt signal integrity. Short lines don't generally require that much attention to termination unless handling very high frequency (short wavelengths relative to cable length) signals.

JR
With input audio devices, however, including outboard line-level gear, you are transferring voltage, not power (look up Ohm's law, voltage dividers). Here, the rule is that the input impedance of the next unit in chain needs to be considerably higher than the generating unit. For example our preamps (Sonic Farm Audio) have a mic input impedance switchable between 1, 2.4 and 10kΩ. There is no perceived level change when switching between them, only with the dynamic mics, whose impedance increases with frequency, there will be a little bit of a high end rolloff at 1k and 2.4k settings, which is the purpose of that switch.

Another example is an audio interface, any brand. As a rule of thumb, the line inputs on those are always solid state, no transformers, and most of them have an input impedance between 5 and 20kΩ. Engineers would go nuts if they had to impedance-match all those line-level units.

Tube stages do have a high output impedance, but the final output of those unit always has a low one, typically below 600Ω. This is achieved either by a step-down transformer or a solid-state buffer. I prefer the second solution, it's more elegant, preserves the tube sound and requires less gain at the tube stage.

One of the most important consideration of mic preamps is noise performance, which has its own impedance matching rules, but this is an issue that designers need to take care of, not the user. Typically, tube and FET stages have a much higher input impedance than transistors, and so mic input transformers in those preamps have a much higher transfer ratio (1:10 or even 1:20). Mic input transformers provide a totally noiseless gain, but they multiply the source (in this case mic) impedance by the square of the transfer ratio, so 100 to 400 times. The "source" impedance seen by the tube/FET will thus be (for a 200Ω mic) 20k or even 80k. Because you can easily achieve a tube/FET stage input impedance of 2.2MΩ or even higher, the signal transfer to the tube will still be optimum. Just don't overdo the transfer ratio, because the input capacitance of the tube/FET stage will start to roll off the high end. There you go, now you can even get into the preamp design.
 
Only power amps and long line drivers need to be impedance-matched to the load, because you want to make sure that maximum power is transferred to the load, which means maximum voltage AND maximum current.
Power amps and long lines are different animals that require different strategies.
Matching source impedance with load impedance results in huge waste of energy, since as much power is dissipated in the source than in the receiver.
The need for matching impedances is to minimize reflections in cables, which is something that happens only when the length of the cable is greater than the signal's wavelength.
For audio signals, it means more than 17 km, about 11 miles. That's why it is considered for telephone landlines.
Speaker lines are typically much shorter, so the main concern is minimizing resistive losses, which happen when the load impedance is smaller than the combined impedance of the source and the wires.
That's the reason why power amp designers try to minimize the output impedance of their amps and installers use the highest possible cable section.
 
In the early days of my apprenticeship at Philips there was a lot of gear around that had high impedance speakers - 600Ω - 1KΩ. These were used in early radiograms and stereograms and there were also 8Ω, 16Ω and 32Ω speakers that had transformers mounted to the basket that ran directly from the tube output stage. Some speaker boxes had transformers in the box - no transformers in the amps.
Philips also developed “motional feedback” speakers in the early ‘70s - they employed a piezo accelerometer mounted on the bass driver dustcap which fed back to the amp to correct LF errors. They were the RH544’s - used on Pink Floyd’s The Wall, The Final Cut, The Division Bell and Roger Waters album Pros &Cons of Hitchhiking.
 
Paralleling output tubes lowers the reflected impedance seen by the load, it does not alter the voltage, much, as global feedback adjust for "missing" output voltage.
The word "match" implies equal values, and as power amps being voltage sources, they ignore this, (in theory) so rather than "matching" impedance I would just call it "suitable" impedance.
Drive an 8 ohm load with a 2 milli Ohm source, no problem. No matching needed :)
 
Interconnect cables don't have a 'characteristic impedance', even if they are hundreds of miles long. Cables only approach their advertised 'Radio Frequency Characteristic Impedance' at frequencies above several hundred kilohertz.
 
That's correct. I remember quite well the amps described in "radio" magazines, that boasted no output xfmr.
hey didn't really catch on, though.
I think if they hung on to the speaker cables they’d catch on. I think Rola and Magnavox and Philips all made high impedance speakers. Would’ve been fun to watch the cone of an 8Ω speaker if you plugged it into the extension speaker jack 🔥 💥.
Remember speakers like these? One good whack and the magnet assembly would come off.
1701803325957.png
 
I vaguely remember philips had 600ohm , or so, voice coil speakers, used in TVs, tabletop radios etc.
 
In the early days of my apprenticeship at Philips there was a lot of gear around that had high impedance speakers - 600Ω - 1KΩ. These were used in early radiograms and stereograms and there were also 8Ω, 16Ω and 32Ω speakers that had transformers mounted to the basket that ran directly from the tube output stage. Some speaker boxes had transformers in the box - no transformers in the amps.
Philips also developed “motional feedback” speakers in the early ‘70s - they employed a piezo accelerometer mounted on the bass driver dustcap which fed back to the amp to correct LF errors. They were the RH544’s - used on Pink Floyd’s The Wall, The Final Cut, The Division Bell and Roger Waters album Pros &Cons of Hitchhiking.
back in the 70s I actually made a road trip down to Arlington VA from CT to do a physical patent search to find and look at that Phillips bass feedback patent and others. This was before we could just let our fingers do the walking on the WWW to search patents.

JR
 
Interconnect cables don't have a 'characteristic impedance',
Yes they do. It doesn't matter in practice, though.
even if they are hundreds of miles long. Cables only approach their advertised 'Radio Frequency Characteristic Impedance' at frequencies above several hundred kilohertz.
Characteristic impedance is unrelated to frequency.
Telephone cables are (were) 600 ohms at vocal frequencies.
 
Last edited:
back in the 70s I actually made a road trip down to Arlington VA from CT to do a physical patent search to find and look at that Phillips bass feedback patent and others. This was before we could just let our fingers do the walking on the WWW to search patents.

JR
Yeah - it’s amazing though, through electronics magazines how quickly word travelled. We were a workshop for Philips warranty and after warranty service and also part of the EDAC and ELA divisions of Philips which involved tech feedback and design testing for tape decks and cassette machines, the first video disc 12” platter machines, Philips proprietary tape noise reduction system DNL (Dynamic Noise Limiter) which they were trying to jam into a car cassette system, Rotor Sound - a solid state stereo Leslie which I think went nowhere, motional feedback speakers and a whole raft of TV innovations. We got a lot of free stuff to “soak test” - run it till it breaks - or not. We got all the latest electronics mags free. Staff price for any components we wanted with no limitations was 20% (1/5) of RRP. We used to get free picture tubes which were rejects if more than 3 phosphors were out (screen fringe only - if centre areas they were rejected with any number). Built my own K9 26” telly in time for the arrival of colour TV first transmission (test) - 1972 Olympics.
 
Back
Top