The Klark Teknik EQP-KT input circuit analysis.

Introduction

The EPQ-KT is a budget copy of the famous Pulse Pultec EQP-1a. The reason why I say copy instead of clone is that a clone is the copy of the exact circuit. I purchase two of these to see how they hold up to the tasks and sounding like the original design. Needless to say, I was disappointed. In this thread, I will be going though my examination technique and show what I do when I analyze a piece of equipment and apply my modding procedures to take a design to the best level of performance. Do not attempt to copy some of my methods without proper bench electronics support equipment and safety practices. Failure to do so might cause yourself to be electrocuted.  

The Pulse Pultec EPQ-1a was created in 1951 for the purpose of putting on the finishing equalization touches of final mixes in the audio production field as well as recorded materials in the broadcast sector. It has been a staple in audio studios for a long time, and now there is even plugins design to try to imitate this piece of equipment. Below we can see the designer's specifications. This was designed to be a broad band equalizer, and using the classic balanced connections that was used in that era.

The Input Circuit
 

Looking at the input circuit, we have a set of inputs, XLR and TRS going into a rfi and an attenuator, then into a transformer on the back of the board. the secondary termination is on the front board.

With injecting a 1V signal, the attenuator knocked it down to 300mV, then stepped the voltage up to 2.5V. Termination of the secondary indicates the output impedance is 10K, so we are looking at a 600 ohm:10K circuit. Rarely will you encounter step up or down transformers with suitable bandwidth response. This transformer, I would use in something narrow band like a microphone preamp where these type of transformers are utilized.

Compare and contrast from the original:

The Triad HS56 is a 600 :600 ohm matching transformer with a frequency response of 10hz-30Khz

The next step is to do signal analysis to determine if we are still within intended specs, even though impedance is different.

Picture 1: The signal at the input connector. Using a scope with a 2.5Vp-p signal @ 50 ohms My signal seems to be loaded down even through a 300 ohm H-pad attenuator. Resistance is low of the two primary coils at 35 ohms where the original was 210 ohms on the primary and secondary and the input impedance was 600 ohms instead of 300 ohms. Some may say "what is the difference". Well output circuits tend to have a higher noise floor when driven down in ohms too much and the bandwidth amplification product of the amp is also effected.   

Now we see what is on the primary of the transformer after the H bridge.

Picture 2: The H-pad attenuated the signal from 2.21V p-p to 1.86V. So, it looks like they built an attenuator so their transformer wouldn't get overloaded.  So what Klark really did was scale down and redesigned this product instead of copying the circuit like how everyone makes a "clone" model. Problem is that changing gain structure here has its negative impact on the circuit, as the current flowing through the primary is lower than the signal at the input connector that would cause a change in noise floor and loss of signal strength as relative core loss will occur with this circuit.

Now, let's look at the Secondary Signal

Picture 3: We see here that our secondary is 1.33V measured without load So the 1:1 transformer in the original design is not being simulated here. What I also find is that the Midas transformer is not 1:1 either, its actually a step up, but they starved the transformer and we are seeing the results. By looking at this transformer and what they were trying to accomplish, is the same voltage level as the input, but did not factor in the losses occurred in the primary. The logical course is to use a more suitable transformer. 

Next slide, we see what the signal is at half of the load of their EQ circuit. (10K resistor across the secondary)

Picture 4. We observe half loading of this starved transformer as we see the voltage dropped again to 1.23V p-p. The reason why I say this is half load is the boost controls are in parallel in Ac signal analysis that have a 10K in impedance. When I measured the current it is only 42.8 uA through the 10K terminator that will be 21.4uA going through the circuit as the series boost controls will half this voltage.

In the original design the terminated at 560 ohms and a higher voltage drop was developed across the 25K boost circuit. Some people said they had limited success increasing the boost resistance of the filters. But that is because this circuit has poor current  delivery, and as a result, the boost starts starving the circuit's overall current and filtering range.

 Replacement transformer suggestions.

I am going to try an Edcor XS4400 transformer. Its a little better designed than the original, 1:1:1:1 quad-filar wound 110 ohms each coil, with a F3 specification of 10Hz-57Khz, +/-1db 20-20000 which is almost exact to a very popular UTC transformer, the A-20 without the encapsulation. At 1/4 the price of the UTC, this sleeper transformer is a modder's dream come true.

The original Triad transformer is a 600 to 600 bifilar wound which isn't currently made, but rumor has it they will be coming back to release that series sometime in 2024. Its a good transformer with a F3 specification (bandwidth measured to -3 db roll off points) is 10hz-30Khz, with a +/- 1db at 20Hz-20KHz. 

Update 01/09/24

Now while I wait for me to get all the stuff together to put the input transformer in a mounting box, mounted on perforated proto board as I wasn't expected having this stuff, so  I had to order it.  

I decided to look ahead to the output circuit.  Another modern circuit that will have to be undone.

Looking at the first tube and drawing the grid circuit. It looks like they didn't terminate the filter, and use a different buffer and went unbalanced into the circuit. This changes a lot of the circuit's function equation, as the filter is not a separate unit from the buffer like in the original design. So the next step will be finding the proper output impedance and terminate with a transformer that can be used as an innerstage. So the 10K resistor they terminated in the front is the filter's average operational impedance which doesn't do an effective job in current distribution in the filter compared to having the input and output terminations that set the operational impedance as the beginning of the filter is low impedance and increases through the filter. 
 

So I have to find out terminating at 420-520 ohms in the beginning will give me the filter and at the end I'll terminate at 100K for testing.