Tube Amps / Music Electronics
For current discussions, please visit Music Electronics Forum.

ampage archive

Vintage threads from the first ten years

Search for:  Mode:  

choosing a PI...

7/3/2005 8:27 PM
swt choosing a PI...
What do you guys prefer, the usual long tail, or a concertina splitter. Sonically speaking...what's the difference between them (cleanest, etc). It's for a small amp, two channels, loop, 2x6v6s. Thanks a lot for your opinions...
And now, a word from our sponsors:

7/6/2005 12:54 PM
steve rathmann
Here's my understanding of some of the diffferences.  
concertina - inherently balanced (up to a point) , easy to to set up, fewer parts, uses a single triode, it however has essentially unity gain, less signal voltage swing, and the outputs are of different impedences which means you can get unbalanced behavior if the output tubes are driven into grid current - however this may not be a bad thing.  
long tail - lots of gain, greater signal swing (headroom), well balanced but may take some tweaking. There is more you CAN do with the LTP by adjusting the plate resistor offset and the "tail". The downside is that it always uses two triodes, more parts and it takes more understanding if you want to dink with the performance.  
Thumbing through any random collection of schematics of years past (from the web or the Tube Amp Book) you'll find that lower powered amps (el84, 6V6) which use cathode biasing often utilize concertina splitters, where the bigger amps with fixed bias went to the long tail. I think this is more than coincidence but certainly not a technical requirement.  
Where cost and simplicity are concerned for a small practice amp the concertina works just fine, but for a higher performance unit where eeking out all possible clean headroom was a priority (remember that originally distortion was unwanted) and could warrant higher cost the LTP was typically used.  
The differences (along with the equivalent difference in power tube setup) are best exemplified sonically by the difference between the "Tweed" sound and the "BF" sound. Neither is better they just have different ideal applications. (Also note though that aside from driver and power tube bias method defferences BF amps also had higher B+ voltages supplied by SS rectifiers with more capacitance.)  
All in all, I originally built my first couple of projects with concertina splitters mostly because I understood how they worked. I didn't quite grok the LTP yet. My last and current project will use the LTP since, as a channel switcher, I want the most out of the clean channel. It won't hurt either that the long tail may drive the 4 6V6s a bit more.
7/6/2005 1:39 PM
Greg Simon
There are also several different paraphase inverter setups that were used in Fender, and several others besides that including the see-saw one mentioned in Kevin O' Connor's TUT series. They will give you less clean headroom and a more unbalanced signal, which can sound great if thats what you're after. One of the best sounding amps I've ever played for a blues lead tone was a '49 Fender Pro, which used a paraphase inverter.  
You'll find that the LTP and concertina can sound light years differently from each other also depending on which tube you use. The 5e5 has a 12ax7 concertina, and the Silvertone 1484 has a 6cg7/6fq7 concertina, and they both sound completely different, with the Silvertone circuit having less gain. The 5f6a bassman uses a 12ax7 LTP, and a blackface Deluxe uses a 12at7, with higher voltages. There are lots of options to choose from here.  
7/7/2005 11:07 AM
steve rathmann
some additional info...
7/6/2005 4:22 PM

My favorite tube crunch tones have come from amps running 2 6V6s with concertina phase splitters.  
7/6/2005 4:59 PM
Dr. Photon

most of the times it's around 2 triodes (1 tube) to get from a vaguely line level signal to to grids of the output tubes. The least amount of gain stages that most small P-P amps need are three. 5E3 deluxe has 4 triodes, two input gain stages mixed together going to a gain stage and split load PI. some Matchless amps have 2 12AX7 - one tube was a gain stage and bias-tremolo osc, the other was a LTP PI.  
a LTP takes two tubes just for the phase inverter, but it has enough gain for the job. The LTP's "balance" can be adjusted by unbalancing the plate loads. It is also capable of more voltage swing (good for big 6L6 or EL84). The LTP provides two inputs, and is fullybalanced. this means that you can make a goofy fully-balanced amp, or provide a balanced line in. some amps can combine two channels into opposite "sides" of the PI (AC-30). NFB is usually applied to the side of the PI opposite to that being driven.  
The paraphase is not often seen, and has fallen out of favorable usage. I think the reason is that it doesn't measure well. I have liked the sound of the few amps I have heard with paraphase inverters. they seem to be almost like single ended designs, with annother stage that slipts off an opposite phase. Most of the designs I have seen use the grid resistors of the output tubes as the feed, and there is much feedback around the inverter stage, to ensure that the gain will always be -1. The preceding gain stage will typically have the feedback applied to its cathode.  
The split load is quite commonly seen. The concertina will typically have a preceeding gain stage, since it has a unity gain. NFB is typically applied to this stage's cathode. I have seen three ways of biasing the concetina stage. The first method is to directly connect it to the plate of the following gain stage. This is typically done in HiFi amps since it takes out one capacitor phase shift from the feedback loop (and saves the cost of a cap). The poblem is that the preceeding stage must be biased correctly. The second method (common in fenders) is to use an appropriate cathode resistor in between the tube's cathode and the resistor to ground. a grid resistor is connected from this point to the tube's grid. I have seen the coupling capacitor connected to the junction of the resistors and also to the cathode - it doesn't seem to matter much. the input coupling capacitor can be small since the grid resistor is essentially bootstrapped (very high input impedance). The last method is to fix the bias for the stage with a pair of resistors. I use resistors in a 2:1 ratio from the plate supply to ground. this method fixes the bias to be independent of tube conditions, or the preceeding stage. Many HiFi amps will use a triode-pentode tube with the direct plate-grid coupling. the pentode as the voltage gain and the triode as the concertina PI. This is used in some ampegs and SUNN amps since their output stage is more like a HiFi amplifier than a regular guitar amp or PA amp is.  
I have a small p-p EL84 stereo amp for my computer speakers that uses the last connection option. I used 2 9-pin triodes for voltage gain and concetina PI. I didn't rig up one tube per channel - I used 1 tube as the volt gain for both stages, and the other as the PI for both channels, since I felt this would lead to more flexibility in tube choice. I first tried connecting the PI grid to the voltage gain stage's plate, but I found this connection to be very dependent upon tube type, and it would only work correctly with the voltage gain stage as a 12AU7. I then added a couplig cap and a 1M/470K resistor to fix the bias on the PI and I can run whatever tubes I like now. I was going to return the NFB to the cathode of the gains stage, but I found that was not necessary with my cathode-feedback output transformers, even with a mushy power supply and pentode connection. great sound!  
annother good choice is an interstage transformer with a centertapped secondary! my current goofey project amp is like this. You can also play with driving a center tapped choke or something.
7/7/2005 3:34 AM
Then there is the cascode PI with gain up to 200-300. Haven't heard anyone using it yet.  
Then you can make a p-p amp with just one pre amp tube. Dual pot for tone and volume.  

  Page 1 of 3 Next> Last Page>>